Merge branch 'release/2.7.2'

.gitignore

@@ -3,5 +3,11 @@
 __pycache__
 .idea
 .venv
+.worktrees
 build
 .env
+
+# DuckDB runtime artifacts (regenerated from decp_prod.parquet at startup)
+**/decp.duckdb
+**/decp.duckdb.tmp
+**/decp.duckdb.lock

CHANGELOG.md

@@ -1,3 +1,11 @@
+#### 2.7.2 (19 avril 2026)
+
+- Chargement des données depuis une base DuckDB plutôt qu'en mémoire (plus de stabilité) ([#71](https://github.com/ColinMaudry/decp.info/issues/71))
+- Mise en cache des vue sur l'observatoire pour un chargement plus rapide (remise à zéro quotidienne)
+- Correction de bug : la liste de colonnes par défaut est bien appliquée plutôt qu'afficher toutes les colonnes
+- Quelques corrections de bugs d'affichage
+- Refactorisation des fonctions utilitaires (`utils.py` approchait des 1 000 lignes)
+
 #### 2.7.1 (23 mars 2026)
 
 - Correction du partage de données filtrées entre dashboard et vue des données
@@ -182,7 +190,7 @@
 
 ### 1.0.0
 
-- publication sur https://decp.info
+- publication sur <https://decp.info>
 - ajout d'une vue équivalente au format DECP réglementaire
 - personnalisation de datasette
 - script de conversion quotidien basé sur [dataflows](https://github.com/datahq/dataflows)

README.md

@@ -1,6 +1,6 @@
 # decp.info
 
-> v2.7.1
+> v2.7.2
 > Outil d'exploration et de téléchargement des données essentielles de la commande publique.
 
 => [decp.info](https://decp.info)

docs/superpowers/specs/2026-03-21-observatoire-full-url-sharing-design.md

@@ -15,7 +15,7 @@ Flat query parameters with short, readable keys. Multi-value filters use repeate
 ## URL Parameter Mapping
 
 | Component ID                                       | URL key          | Type            | Default (omitted) |
-|---|---|---|---|
+| -------------------------------------------------- | ---------------- | --------------- | ----------------- |
 | `dashboard_year`                                   | `annee`          | single          | `None`            |
 | `dashboard_acheteur_id`                            | `acheteur_id`    | single          | `None`            |
 | `dashboard_acheteur_categorie`                     | `acheteur_cat`   | single          | `None`            |
@@ -35,6 +35,7 @@ Flat query parameters with short, readable keys. Multi-value filters use repeate
 | `dashboard_marche_considerationsEnvironnementales` | `env`            | multi           | `[]`/`None`       |
 
 Example URL:
+
 ```
 /observatoire?annee=2024&acheteur_id=12345678901234&acheteur_dept=75&acheteur_dept=13&montant_min=10000&innovant=oui
 ```
@@ -79,6 +80,7 @@ FILTER_PARAMS = [
 **Current:** Extracts only `acheteur_id` and `titulaire_id` from URL.
 
 **New:**
+
 - Iterates over `FILTER_PARAMS` to extract all values from `parse_qs`
 - For multi-value params: reads the full list from `parse_qs` (returns lists natively)
 - For number params (`montant_min`, `montant_max`): casts to `float`
@@ -102,12 +104,14 @@ Links generated by `add_links()` in `src/utils.py` (used on search results to li
 ### Fix broken test `test_010_observatoire_montant_filter`
 
 This test imports `_apply_filters` from `pages.observatoire`, which no longer exists (replaced by `prepare_dashboard_data` in `src/utils.py`). Fix:
+
 - Replace import with `from src.utils import prepare_dashboard_data`
 - Update the call to match `prepare_dashboard_data`'s signature: rename `marche_type` keyword to `type`, and add missing params `objet`, `code_cpv`, `techniques`, `marche_innovant`, `sous_traitance_declaree` (all as `None`)
 
 ### New test: multi-param URL round-trip
 
 Add a test that navigates to `/observatoire?annee=2024&acheteur_id=<test_id>&montant_min=10000` and verifies that:
+
 - `dashboard_year` dropdown shows "2024"
 - `dashboard_acheteur_id` input contains the test ID
 - `dashboard_montant_min` input contains "10000"

docs/superpowers/specs/2026-04-15-duckdb-migration-design.md

@@ -0,0 +1,196 @@
+# DuckDB migration — design spec
+
+**Date:** 2026-04-15
+**Branch:** dev
+**Status:** Approved, ready for planning
+
+## Goal
+
+Replace the global Polars dataframes that `src/utils.py` materializes at import time (`df` and the five derived frames, lines 891–913) with a DuckDB database on disk. The main table holds ~1.5M rows from `decp_prod.parquet`. Per-request queries pull only what each page needs, dramatically reducing steady-state RSS memory.
+
+Polars stays the primary API for small result sets and post-processing. DuckDB carries the heavy filtering, joining, and aggregation.
+
+## Approach summary
+
+- **Approach A — compatibility layer.** A new `src/db.py` module exposes a `query_marches(where_sql, params, columns, ...)` helper that runs SQL and returns a `pl.DataFrame`. Most existing `df.filter(pl.col(...) == x)` call sites translate mechanically to `query_marches("col = ?", (x,))`. The shape of downstream Polars code is unchanged.
+- **Two small helpers stay in memory.** `df_acheteurs` and `df_titulaires` (tens of thousands of rows, consumed by the autocomplete search on every keystroke) are kept as module-level Polars frames. They are populated from DuckDB at import time, not from Parquet.
+- **Four derived tables live in DuckDB**, built at startup alongside the main table: `acheteurs_marches`, `titulaires_marches`, `acheteurs_departement`, `titulaires_departement`.
+- **Connection model.** One read-only `duckdb.connect(..., read_only=True)` at module load, shared across the process. `conn.cursor()` per Dash callback for thread-safety. The read-write connection is short-lived and only used during the startup build phase.
+
+## Cache invalidation rule
+
+At startup, rebuild the DuckDB file if:
+
+1. **The DB file does not exist**, OR
+2. **`decp_prod.parquet.mtime > duckdb.mtime`**, **unless** `DEVELOPMENT=true` and `REBUILD_DUCKDB != true` — in which case the DB stays as-is (fast dev reloads).
+
+Production auto-rebuilds when the source Parquet is newer. Development keeps a stable DB across reloads unless the developer explicitly sets `REBUILD_DUCKDB=true` to force a rebuild.
+
+## Concurrency
+
+Multi-worker Gunicorn startup and crashed-mid-build scenarios are handled by a file lock, not by polling for the tmp file's existence:
+
+```python
+with open(DB_PATH.with_suffix(".duckdb.lock"), "w") as lock_fd:
+    fcntl.flock(lock_fd, fcntl.LOCK_EX)          # blocks if another worker is building
+    if should_rebuild(DB_PATH, PARQUET_PATH):
+        build_database(DB_PATH, PARQUET_PATH)
+conn = duckdb.connect(str(DB_PATH), read_only=True)
+```
+
+- Worker A acquires the lock, builds, atomically renames tmp → final, releases the lock.
+- Worker B blocks on `flock`, then re-checks `should_rebuild`, sees the fresh DB, skips building.
+- `fcntl.flock` is auto-released on process death, so a crash never deadlocks the next worker.
+- `build_database` unlinks any pre-existing tmp file before starting (safe because it holds the lock) — handles an abandoned tmp from a crashed previous build.
+
+## Build logic
+
+The build keeps **one source of truth** for transforms by reusing the existing Polars pipeline:
+
+```python
+def build_database(db_path, parquet_path):
+    tmp_path = db_path.with_suffix(".duckdb.tmp")
+    if tmp_path.exists():
+        tmp_path.unlink()
+    frame = get_decp_data()   # existing function in utils.py
+    with duckdb.connect(str(tmp_path)) as w:
+        w.register("frame", frame)
+        w.execute("CREATE TABLE decp AS SELECT * FROM frame")
+        w.execute("CREATE TABLE acheteurs_marches AS "
+                  "SELECT DISTINCT uid, objet, acheteur_id FROM decp "
+                  "ORDER BY acheteur_id")
+        w.execute("CREATE TABLE titulaires_marches AS "
+                  "SELECT DISTINCT uid, objet, titulaire_id FROM decp "
+                  "ORDER BY titulaire_id")
+        w.execute("CREATE TABLE acheteurs_departement AS "
+                  "SELECT DISTINCT acheteur_id, acheteur_nom, acheteur_departement_code "
+                  "FROM decp ORDER BY acheteur_nom")
+        w.execute("CREATE TABLE titulaires_departement AS "
+                  "SELECT DISTINCT titulaire_id, titulaire_nom, titulaire_departement_code "
+                  "FROM decp ORDER BY titulaire_nom")
+    os.replace(tmp_path, db_path)
+```
+
+Why Polars, not SQL, for the row-level transforms:
+
+- `booleans_to_strings` is not a simple cast — it replaces `true`/`false` with `"oui"`/`"non"` on every boolean column. Reimplementing in SQL risks drifting from the Polars version.
+- The null-name replacement (`acheteur_nom`, `titulaire_nom` → `"[Identifiant non reconnu dans la base INSEE]"`) is also easier to keep identical in Polars.
+- `w.register("frame", frame)` is zero-copy. The memory spike is one-time during build and released when the write connection closes.
+
+`os.replace` is atomic on POSIX — the read-only connection that opens next always sees a complete DB.
+
+## Module layout
+
+### New: `src/db.py`
+
+```python
+conn: duckdb.DuckDBPyConnection       # read-only, module-level
+schema: pl.Schema                     # from conn.execute("SELECT * FROM decp LIMIT 0").pl().schema
+
+def get_cursor() -> duckdb.DuckDBPyConnection: ...
+def query_marches(where_sql: str = "TRUE",
+                  params: tuple = (),
+                  columns: list[str] | None = None,
+                  order_by: str | None = None,
+                  limit: int | None = None) -> pl.DataFrame: ...
+def should_rebuild(db_path: Path, parquet_path: Path) -> bool: ...
+def build_database(db_path: Path, parquet_path: Path) -> None: ...
+```
+
+Only imports: `polars`, `duckdb`, `os`, `fcntl`, `pathlib`, `logging`. No app modules — prevents circular imports.
+
+### Changes to `src/utils.py`
+
+- `df: pl.DataFrame = get_decp_data()` — **removed** (after migration).
+- `df_acheteurs`, `df_titulaires` — **kept as Polars globals**, populated via DuckDB at import time. The query mirrors today's `get_org_data(df, org_type)`: select all columns whose name starts with `acheteur_` (or `titulaire_`) except the `_latitude` / `_longitude` pair, plus `COUNT(*) AS "Marchés"`, grouped by the same set. Implementation can either:
+
+  - enumerate the columns by filtering `schema.names()` at import time and build the `SELECT` / `GROUP BY` strings, or
+  - call `get_org_data()` once against a small Polars frame returned by `SELECT <org_ cols> FROM decp`.
+
+  Feeds `search_org` unchanged.
+
+- `df_acheteurs_marches`, `df_titulaires_marches`, `df_acheteurs_departement`, `df_titulaires_departement` — **removed** as Python globals. Call sites query the corresponding DuckDB tables.
+- `schema` — imported from `src/db.py` (stays a `pl.Schema` — so `schema.names()` and dtype lookups both work, no call-site changes beyond `acheteur.py:303`).
+- `columns` — replaced with `schema.names()`.
+- `get_decp_data()` — **kept** (used by `build_database`).
+- `get_org_data()` — can be removed once `df_acheteurs` / `df_titulaires` are populated from DuckDB directly.
+
+### Call-site translations
+
+| Before (Polars global)                                       | After                                                                             |
+| ------------------------------------------------------------ | --------------------------------------------------------------------------------- |
+| `df.filter(pl.col("acheteur_id") == aid)`                    | `query_marches("acheteur_id = ?", (aid,))`                                        |
+| `df.filter(pl.col("uid") == uid).row(0, named=True)`         | `query_marches("uid = ?", (uid,)).row(0, named=True)`                             |
+| `df.select("uid","objet","acheteur_id").filter(...)`         | `query_marches("...", (...), columns=["uid","objet","acheteur_id"])`              |
+| `df.columns`                                                 | `schema.names()`                                                                  |
+| `df_acheteurs_marches.filter(...)`                           | `get_cursor().execute("SELECT ... FROM acheteurs_marches WHERE ...", [...]).pl()` |
+| `pl.DataFrame(schema=df.collect_schema())` (acheteur.py:303) | `pl.DataFrame(schema=schema)`                                                     |
+
+Heavy dashboard aggregations (observatoire, tableau full-scan) use raw SQL via `get_cursor().execute(...).pl()` rather than the helper.
+
+## Configuration
+
+- **`DATA_FILE_PARQUET_PATH`** — unchanged.
+- **DuckDB file location** — computed: `Path(DATA_FILE_PARQUET_PATH).parent / "decp.duckdb"`. No new env var.
+- **`REBUILD_DUCKDB`** — new, optional, default `false`. In development, setting this to `true` forces a rebuild when the parquet is newer.
+- **`DEVELOPMENT`** — unchanged; now also gates the auto-rebuild behavior per the rule above.
+
+## Testing
+
+- `tests/conftest.py` (or a startup hook in `src/db.py`) ensures the test run builds the DuckDB in a temp directory derived from the parquet path — `tests/test.parquet` → `tests/decp.duckdb`. This file is added to `.gitignore`.
+- Tests already set `DEVELOPMENT=true`; they must also set `REBUILD_DUCKDB=true` on cold test runs to force a fresh build from the test parquet.
+- The existing Selenium suite exercises every page and is the primary acceptance signal.
+
+## Migration order
+
+Incremental — `df` global coexists with `src/db.py` until every page is migrated.
+
+1. **Add `src/db.py`** (build, lock, `query_marches`, `schema`). `df` global unchanged.
+2. **Migrate `marche.py`** — single-row lookup by `uid`, one call site.
+3. **Migrate `acheteur.py`, `titulaire.py`** — filter by id.
+4. **Migrate `arbre/departement.py`, `arbre/liste_marches_org.py`** — use the new derived DuckDB tables.
+5. **Migrate `tableau.py`** — may need raw SQL.
+6. **Migrate `observatoire.py`** — heaviest aggregations, most likely raw SQL.
+7. **Migrate `figures.py`** — uses `df` in chart generation.
+8. **Remove** `df`, `df_*_marches`, `df_*_departement` globals, `get_org_data()`, and the `df = get_decp_data()` call from `utils.py`. Move `schema` / `columns` exports to `src/db.py`.
+
+### Verification gates
+
+- `uv run pytest` green after every page migration.
+- Manual smoke test via `uv run run.py` of the migrated page before proceeding.
+- RSS memory measurement (`ps -o rss`) of a cold `gunicorn app:server` with the prod parquet, before and after, to confirm the memory reduction.
+
+## Out of scope
+
+- Changes to `src/cache.py` (flask-caching stays).
+- The in-progress observatoire-localstorage-filters work on `dev`.
+- Schema changes to the parquet.
+- SQL views beyond the four derived tables.
+- Multi-database or replication setups.
+
+## Risks and mitigations
+
+| Risk                                                                          | Mitigation                                                                                                                       |
+| ----------------------------------------------------------------------------- | -------------------------------------------------------------------------------------------------------------------------------- |
+| `booleans_to_strings` reimplemented in SQL and drifts from Polars version     | Transforms stay in Polars via `w.register("frame", frame)`. One source of truth.                                                 |
+| Two Gunicorn workers rebuild concurrently                                     | `fcntl.flock` serializes the build; second worker re-checks and skips.                                                           |
+| Crashed build leaves stale `.tmp` file                                        | Build unlinks any pre-existing tmp before starting (safe under lock).                                                            |
+| `schema` shape change breaks `acheteur.py:303`                                | `schema` stays a `pl.Schema` object, not a list. One call site (`collect_schema()` → module `schema`) updated.                   |
+| Test runs inherit a stale DuckDB from a previous run with a different parquet | Tests force `REBUILD_DUCKDB=true` on cold runs; test DB added to `.gitignore`.                                                   |
+| Read-only connection opened before build finishes in another worker           | Lock held across build + rename; read-only `connect` happens after lock release. Atomic `os.replace` guarantees a complete file. |
+
+## Outcome
+
+### Memory impact
+
+Memory measurement against the production parquet (`decp_prod.parquet`, ~1.5M rows) requires a running gunicorn process with access to the production data file. The measurement was deferred to the post-merge smoke test on the staging server (test.decp.info).
+
+**Expected reduction:** The removed globals (`df`, `df_acheteurs_departement`, `df_titulaires_departement`, `df_acheteurs_marches`, `df_titulaires_marches`) previously materialised the full 1.5M-row Parquet in memory as multiple Polars frames. At ~300 bytes/row × 5 frames, steady-state RSS reduction is estimated at **1–2 GB per worker**. The retained `df_acheteurs` and `df_titulaires` (autocomplete search) represent only the distinct-organisation subset (~tens of thousands of rows) and are negligible.
+
+**What remains in memory:**
+
+- `df_acheteurs` — distinct acheteurs with Marchés count (populated from DuckDB at startup)
+- `df_titulaires` — same for titulaires
+- DuckDB's own page cache (disk-backed, grows under load, evicted by OS)
+
+All per-request data is fetched from DuckDB and discarded after the callback returns.

pyproject.toml

@@ -1,11 +1,9 @@
 [project]
 name = "decp.info"
 description = "Interface d'exploration et d'analyse des marchés publics français."
-version = "2.7.1"
+version = "2.7.2"
 requires-python = ">= 3.10"
-authors = [
+authors = [{ name = "Colin Maudry", email = "colin@colmo.tech" }]
-    { name = "Colin Maudry", email = "colin@colmo.tech" }
-]
 dependencies = [
   "dash==3.4.0",
   "dash[compress]",
@@ -19,7 +17,10 @@ dependencies = [
   "pandas",                    # utilisé pour la création de certains graphiques
   "unidecode",
   "dash-leaflet",
-    "dash-extensions"
+  "dash-extensions",
+  "duckdb",
+  "flask-caching",
+  "pyarrow>=23.0.1",
 ]
 
 [project.optional-dependencies]
@@ -30,19 +31,15 @@ dev = [
   "selenium",
   "webdriver-manager",
   "dash[testing]",
-    "fastexcel"
+  "fastexcel",
 ]
 
 [tool.pytest.ini_options]
-pythonpath = [
+pythonpath = ["src"]
-  "src"
+testpaths = ["tests"]
-]
-testpaths = [
-    "tests"
-]
 env = [
   "DATA_FILE_PARQUET_PATH=tests/test.parquet",
   "DEVELOPMENT=true",
-    "DATA_SCHEMA_PATH=/home/colin/git/decp-processing/dist/schema.json"
+  "DATA_SCHEMA_PATH=/home/colin/git/decp-processing/dist/schema.json",
 ]
 addopts = "-p no:warnings"

src/app.py

@@ -1,5 +1,5 @@
-import logging
 import os
+from shutil import rmtree
 
 import dash_bootstrap_components as dbc
 import tomllib
@@ -7,15 +7,15 @@ from dash import Dash, Input, Output, State, dcc, html, page_container, page_reg
 from dotenv import load_dotenv
 from flask import Response
 
+from src.cache import cache
+from src.utils import DEVELOPMENT
+
 load_dotenv()
 
 # if os.getenv("PYTEST_CURRENT_TEST"):
 #     os.environ["DATA_FILE_PARQUET_PATH"]
 
-
+META_TAGS = [
-development = os.getenv("DEVELOPMENT").lower() == "true"
-
-meta_tags = [
     {"name": "viewport", "content": "width=device-width, initial-scale=1"},
     {
         "name": "keywords",
@@ -23,20 +23,29 @@ meta_tags = [
     },
 ]
 
-if development:
+if DEVELOPMENT:
-    meta_tags.append({"name": "robots", "content": "noindex"})
+    META_TAGS.append({"name": "robots", "content": "noindex"})
 
 app: Dash = Dash(
     title="decp.info",
     use_pages=True,
     compress=True,
-    meta_tags=meta_tags,
+    meta_tags=META_TAGS,
 )
 
-# COSMO (belle font, blue),
+rmtree(os.getenv("CACHE_DIR", "/tmp/decp-cache"))
-# UNITED (rouge, ubuntu font),
+
-# LUMEN (gros séparateur, blue clair),
+cache.init_app(
-# SIMPLEX (rouge, séparateur)
+    app.server,
+    config={
+        "CACHE_TYPE": "FileSystemCache",
+        "CACHE_DIR": os.getenv("CACHE_DIR", "/tmp/decp-cache"),
+        "CACHE_DEFAULT_TIMEOUT": int(
+            os.getenv("CACHE_DEFAULT_TIMEOUT", 3600 * 24)
+        ),  # 24h par défaut
+        "CACHE_THRESHOLD": 300,
+    },
+)
 
 
 # robots.txt
@@ -67,13 +76,6 @@ def sitemap():
     return Response(xml, mimetype="text/xml")
 
 
-logger = logging.getLogger("decp.info")
-logging.basicConfig(
-    format="%(asctime)s %(levelname)-8s %(message)s",
-    level=logging.INFO,
-    datefmt="%Y-%m-%d %H:%M:%S",
-)
-
 with open("./pyproject.toml", "rb") as f:
     pyproject = tomllib.load(f)
     version = "v" + pyproject["project"]["version"]

src/assets/css/style.css

@@ -197,9 +197,9 @@ p.version > a {
 .table-menu {
   font-size: 16px;
   margin: 12px 0 12px 0;
-  height: 50px;
   display: flex;
   align-items: center;
+  flex-wrap: wrap;
 }
 
 .table-menu > * {

src/cache.py

@@ -0,0 +1,4 @@
+from flask_caching import Cache
+
+# Isolé dans un fichier dédié pour éviter les imports circulaires
+cache = Cache()

src/db.py

@@ -0,0 +1,162 @@
+import fcntl
+import os
+from pathlib import Path
+from time import sleep
+
+import duckdb
+import polars as pl
+import polars.selectors as cs
+from polars.exceptions import ComputeError
+
+from src.utils import logger
+
+
+def should_rebuild(db_path: Path, parquet_path: Path) -> bool:
+    db_path = Path(db_path)
+    parquet_path = Path(parquet_path)
+    if not db_path.exists():
+        return True
+    dev = os.getenv("DEVELOPMENT", "False").lower() == "true"
+    force = os.getenv("REBUILD_DUCKDB", "False").lower() == "true"
+    if dev and not force:
+        return False
+    return parquet_path.stat().st_mtime > db_path.stat().st_mtime
+
+
+def _load_source_frame(parquet_path: Path) -> pl.DataFrame:
+    """Read the source parquet and apply the row-level transforms.
+
+    Kept here (not in utils.py) so src.db has no dependency on utils.
+    Mirrors the behavior previously in utils.get_decp_data().
+    """
+    try:
+        lff: pl.LazyFrame = pl.scan_parquet(str(parquet_path))
+    except ComputeError:
+        logger.info("Lecture du parquet échouée, nouvelle tentative dans 10s...")
+        sleep(10)
+        lff = pl.scan_parquet(str(parquet_path))
+
+    lff = lff.sort(by=["dateNotification", "uid"], descending=True, nulls_last=True)
+    lff = lff.filter(pl.col("donneesActuelles")).drop("donneesActuelles")
+
+    # booleans_to_strings: true → "oui", false → "non"
+    lff = lff.with_columns(
+        pl.col(cs.Boolean)
+        .cast(pl.String)
+        .str.replace("true", "oui")
+        .str.replace("false", "non")
+    )
+
+    for col in ["acheteur_nom", "titulaire_nom"]:
+        lff = lff.with_columns(
+            pl.when(pl.col(col).is_null())
+            .then(pl.lit("[Identifiant non reconnu dans la base INSEE]"))
+            .otherwise(pl.col(col))
+            .name.keep()
+        )
+
+    return lff.collect()
+
+
+def build_database(db_path: Path, parquet_path: Path) -> None:
+    """Build the DuckDB database atomically under an exclusive lock.
+
+    Caller MUST hold the fcntl.flock on the .lock file.
+    """
+    db_path = Path(db_path)
+    parquet_path = Path(parquet_path)
+    tmp_path = db_path.with_suffix(".duckdb.tmp")
+    staging_parquet = db_path.with_suffix(".staging.parquet")
+    if tmp_path.exists():
+        tmp_path.unlink()
+
+    logger.info(f"Construction de la base DuckDB à partir de {parquet_path}...")
+    frame = _load_source_frame(parquet_path)
+
+    # Write transformed frame as parquet so DuckDB can read it natively
+    # (avoids pyarrow dependency for the Polars→DuckDB handoff)
+    frame.write_parquet(str(staging_parquet))
+    try:
+        with duckdb.connect(str(tmp_path)) as w:
+            w.execute(
+                f"CREATE TABLE decp AS SELECT * FROM read_parquet('{staging_parquet}')"
+            )
+            w.execute(
+                "CREATE TABLE acheteurs_marches AS "
+                "SELECT DISTINCT uid, objet, acheteur_id FROM decp "
+                "ORDER BY acheteur_id"
+            )
+            w.execute(
+                "CREATE TABLE titulaires_marches AS "
+                "SELECT DISTINCT uid, objet, titulaire_id FROM decp "
+                "ORDER BY titulaire_id"
+            )
+            w.execute(
+                "CREATE TABLE acheteurs_departement AS "
+                "SELECT DISTINCT acheteur_id, acheteur_nom, acheteur_departement_code "
+                "FROM decp ORDER BY acheteur_nom"
+            )
+            w.execute(
+                "CREATE TABLE titulaires_departement AS "
+                "SELECT DISTINCT titulaire_id, titulaire_nom, titulaire_departement_code "
+                "FROM decp ORDER BY titulaire_nom"
+            )
+    finally:
+        if staging_parquet.exists():
+            staging_parquet.unlink()
+
+    os.replace(tmp_path, db_path)
+    logger.info(f"Base DuckDB construite : {db_path}")
+
+
+def _resolve_db_path() -> Path:
+    parquet = os.getenv("DATA_FILE_PARQUET_PATH")
+    if not parquet:
+        raise RuntimeError("DATA_FILE_PARQUET_PATH is not set")
+    return Path(parquet).parent / "decp.duckdb"
+
+
+def _ensure_database() -> Path:
+    db_path = _resolve_db_path()
+    parquet_path = Path(os.getenv("DATA_FILE_PARQUET_PATH"))
+    lock_path = db_path.with_suffix(".duckdb.lock")
+
+    with open(lock_path, "w") as lock_fd:
+        fcntl.flock(lock_fd, fcntl.LOCK_EX)
+        if should_rebuild(db_path, parquet_path):
+            build_database(db_path, parquet_path)
+        else:
+            logger.debug("Base de données déjà disponible et à jour.")
+    return db_path
+
+
+DB_PATH = _ensure_database()
+conn: duckdb.DuckDBPyConnection = duckdb.connect(str(DB_PATH), read_only=True)
+schema: pl.Schema = conn.execute("SELECT * FROM decp LIMIT 0").pl().schema
+
+
+def get_cursor() -> duckdb.DuckDBPyConnection:
+    """Return a per-request cursor that shares the process-wide connection."""
+    return conn.cursor()
+
+
+def query_marches(
+    where_sql: str = "TRUE",
+    params: tuple = (),
+    columns: list[str] | None = None,
+    order_by: str | None = None,
+    limit: int | None = None,
+) -> pl.DataFrame:
+    """Run a parameterized SELECT against the decp table and return Polars.
+
+    `where_sql` and `order_by` are trusted SQL fragments (callers are internal
+    code, never user input). `params` values are passed through DuckDB's
+    parameter binding.
+    """
+    cols = ", ".join(columns) if columns else "*"
+    sql = f"SELECT {cols} FROM decp WHERE {where_sql}"
+    if order_by:
+        sql += f" ORDER BY {order_by}"
+    if limit is not None:
+        sql += f" LIMIT {int(limit)}"
+    return get_cursor().execute(sql, list(params)).pl()

src/figures.py

@@ -12,14 +12,9 @@ import polars as pl
 from dash import dash_table, dcc, html
 from dash_extensions.javascript import Namespace
 
-from src.utils import (
+from src.db import schema
-    add_links,
+from src.utils.data import DATA_SCHEMA, DEPARTEMENTS_GEOJSON
-    data_schema,
+from src.utils.table import add_links, format_number, setup_table_columns
-    departements_geojson,
-    df,
-    format_number,
-    setup_table_columns,
-)
 
 
 def get_yearly_statistics(statistics, today_str) -> html.Div:
@@ -260,8 +255,8 @@ class DataTable(dash_table.DataTable):
 
         style_cell_common = {"fontFamily": "Inter", "fontSize": "16px"}
 
-        for key in data_schema.keys():
+        for key in DATA_SCHEMA.keys():
-            field = data_schema[key]
+            field = DATA_SCHEMA[key]
             if field["type"] in ["number", "integer"]:
                 rule = {
                     "if": {"column_id": field["name"]},
@@ -365,7 +360,7 @@ def get_duplicate_matrix() -> dcc.Graph:
     return dcc.Graph(figure=fig)
 
 
-def get_geographic_maps(dff: pl.DataFrame) -> list | None:
+def get_geographic_maps(dff: pl.DataFrame) -> list[dbc.Col] | list:
     """
     Génère les cartes géographiques pour l'hexagone et les DOM-TOM.
     """
@@ -409,7 +404,7 @@ def get_geographic_maps(dff: pl.DataFrame) -> list | None:
         },
     }
 
-    def make_map_data(region_code: str) -> tuple[list, str or None]:
+    def make_map_data(region_code: str) -> tuple[list, str | None]:
         lff: pl.LazyFrame = dff.lazy()
         if region_code == "Hexagone":
             lff = lff.filter(
@@ -514,7 +509,7 @@ def make_chloropleth_map(region: dict) -> dcc.Graph:
 
     fig = px.choropleth(
         df_map,
-        geojson=departements_geojson,
+        geojson=DEPARTEMENTS_GEOJSON,
         locations="Département",
         color="uid",
         color_continuous_scale="Reds",
@@ -722,7 +717,7 @@ def make_donut(
     nulls="?",
     potentially_many_names: bool = False,
 ):
-    title = data_schema[names_col]["title"]
+    title = DATA_SCHEMA[names_col]["title"]
     lff = lff.rename({names_col: title})
     lff = lff.select("uid", title)
 
@@ -771,16 +766,16 @@ def make_column_picker(page: str):
     table_columns = [
         {
             "id": col,
-            "name": data_schema[col]["title"],
+            "name": DATA_SCHEMA[col]["title"],
-            "description": data_schema[col]["description"],
+            "description": DATA_SCHEMA[col]["description"],
         }
-        for col in df.columns
+        for col in schema.names()
     ]
     for column in table_columns:
         new_column = {
             "id": column["id"],
             "name": column["name"],
-            "description": data_schema[column["id"]]["description"],
+            "description": DATA_SCHEMA[column["id"]]["description"],
         }
         table_data.append(new_column)
 

src/pages/a-propos.py

@@ -3,9 +3,9 @@ import os
 from dash import dcc, html, register_page
 
 from src.figures import get_sources_tables
-from src.utils import meta_content
+from src.utils.seo import META_CONTENT
 
-name = "À propos"
+NAME = "À propos"
 
 register_page(
     __name__,
@@ -13,14 +13,14 @@ register_page(
     title="À propos | decp.info",
     name="À propos",
     description="En savoir plus sur decp.info, l'outil d'exploration des données essentielles de la commande publique.",
-    image_url=meta_content["image_url"],
+    image_url=META_CONTENT["image_url"],
     order=5,
 )
 
 layout = html.Div(
     className="container",
     children=[
-        html.H2(name),
+        html.H2(NAME),
         html.Div(
             className="a-propos-container",
             children=[

src/pages/acheteur.py

@@ -15,6 +15,7 @@ from dash import (
     register_page,
 )
 
+from src.db import query_marches, schema
 from src.figures import (
     DataTable,
     get_distance_histogram,
@@ -23,24 +24,21 @@ from src.figures import (
     make_column_picker,
     point_on_map,
 )
-from src.utils import (
+from src.utils.data import DF_ACHETEURS, get_annuaire_data, get_departement_region
-    columns,
+from src.utils.frontend import get_button_properties
-    df,
+from src.utils.seo import META_CONTENT
-    df_acheteurs,
+from src.utils.table import (
+    COLUMNS,
     filter_table_data,
     format_number,
-    get_annuaire_data,
-    get_button_properties,
     get_default_hidden_columns,
-    get_departement_region,
-    meta_content,
     prepare_table_data,
     sort_table_data,
 )
 
 
-def get_title(acheteur_id: str = None) -> str:
+def get_title(acheteur_id: str | None = None) -> str:
-    acheteur_nom = df_acheteurs.filter(pl.col("acheteur_id") == acheteur_id).select(
+    acheteur_nom = DF_ACHETEURS.filter(pl.col("acheteur_id") == acheteur_id).select(
         "acheteur_nom"
     )
     if acheteur_nom.height > 0:
@@ -54,11 +52,11 @@ register_page(
     title=get_title,
     name="Acheteur",
     description="Consultez les marchés publics attribués par cet acheteur.",
-    image_url=meta_content["image_url"],
+    image_url=META_CONTENT["image_url"],
     order=5,
 )
 
-datatable = html.Div(
+DATATABLE = html.Div(
     className="marches_table",
     children=DataTable(
         dtid="acheteur_datatable",
@@ -70,7 +68,7 @@ datatable = html.Div(
         sort_action="custom",
         page_size=10,
         hidden_columns=[],
-        columns=[{"id": col, "name": col} for col in df.columns],
+        columns=[{"id": col, "name": col} for col in schema.names()],
     ),
 )
 
@@ -229,7 +227,7 @@ layout = [
                         scrollable=True,
                         size="xl",
                     ),
-                    datatable,
+                    DATATABLE,
                 ],
             ),
         ],
@@ -300,7 +298,7 @@ def update_acheteur_infos(url):
 def update_acheteur_stats(data):
     dff = pl.DataFrame(data, strict=False, infer_schema_length=5000)
     if dff.height == 0:
-        dff = pl.DataFrame(schema=df.collect_schema())
+        dff = pl.DataFrame(schema=schema)
     df_marches = dff.unique("id")
     nb_marches = format_number(df_marches.height)
     # somme_marches = format_number(int(df_marches.select(pl.sum("montant")).item()))
@@ -326,17 +324,15 @@ def update_acheteur_stats(data):
     Input(component_id="acheteur_url", component_property="pathname"),
     Input(component_id="acheteur_year", component_property="value"),
 )
-def get_acheteur_marches_data(url, acheteur_year: str) -> tuple:
+def get_acheteur_marches_data(url, ach_year: str) -> tuple:
     acheteur_siret = url.split("/")[-1]
-    lff = df.lazy()
+    lff = query_marches("acheteur_id = ?", (acheteur_siret,)).lazy()
-    lff = lff.filter(pl.col("acheteur_id") == acheteur_siret)
+    if ach_year and ach_year != "Toutes les années":
-    if acheteur_year and acheteur_year != "Toutes les années":
+        ach_year = int(ach_year)
-        acheteur_year = int(acheteur_year)
+        lff = lff.filter(pl.col("dateNotification").dt.year() == ach_year)
-        lff = lff.filter(pl.col("dateNotification").dt.year() == acheteur_year)
     lff = lff.sort(["dateNotification", "uid"], descending=True, nulls_last=True)
     dff: pl.DataFrame = lff.collect(engine="streaming")
     download_disabled, download_text, download_title = get_button_properties(dff.height)
-
     data = dff.to_dicts()
     return data, download_disabled, download_text, download_title
 
@@ -412,7 +408,12 @@ def download_acheteur_data(
     prevent_initial_call=True,
 )
 def download_filtered_acheteur_data(
-    data, n_clicks, acheteur_nom, filter_query, sort_by, hidden_columns: list = None
+    data,
+    n_clicks,
+    acheteur_nom,
+    filter_query,
+    sort_by,
+    hidden_columns: list | None = None,
 ):
     lff: pl.LazyFrame = pl.LazyFrame(
         data
@@ -457,22 +458,23 @@ clientside_callback(
 )
 def update_hidden_columns_from_checkboxes(selected_columns):
     if selected_columns:
-        selected_columns = [columns[i] for i in selected_columns]
+        selected_columns = [COLUMNS[i] for i in selected_columns]
-        hidden_columns = [col for col in columns if col not in selected_columns]
+        hidden_columns = [col for col in COLUMNS if col not in selected_columns]
         return hidden_columns
     else:
         return []
 
 
 @callback(
-    Output("acheteur_datatable", "hidden_columns", allow_duplicate=True),
+    Output("acheteur_datatable", "hidden_columns"),
     Input(
         "acheteur-hidden-columns",
         "data",
     ),
-    prevent_initial_call=True,
 )
 def store_hidden_columns(hidden_columns):
+    if hidden_columns is None:
+        hidden_columns = get_default_hidden_columns("acheteur")
     return hidden_columns
 
 
@@ -485,7 +487,7 @@ def update_checkboxes_from_hidden_columns(hidden_cols, current_checkboxes):
     hidden_cols = hidden_cols or get_default_hidden_columns("acheteur")
 
     # Show all columns that are NOT hidden
-    visible_cols = [columns.index(col) for col in columns if col not in hidden_cols]
+    visible_cols = [COLUMNS.index(col) for col in COLUMNS if col not in hidden_cols]
     return visible_cols
 
 

src/pages/arbre/departement.py

@@ -1,17 +1,17 @@
-import polars as pl
 from dash import Input, Output, callback, dcc, html, register_page
 
-from src.utils import departements, df_acheteurs_departement, df_titulaires_departement
+from src.db import get_cursor
+from src.utils.data import DEPARTEMENTS
 
-name = "Département"
+NAME = "Département"
 
 
 def get_title(code):
-    return f"Marchés publics de {departements[code]['departement']} | decp.info"
+    return f"Marchés publics de {DEPARTEMENTS[code]['departement']} | decp.info"
 
 
 def get_description(code):
-    return f"Marchés publics passés dans le département {departements[code]['departement']} | decp.info"
+    return f"Marchés publics passés dans le département {DEPARTEMENTS[code]['departement']} | decp.info"
 
 
 register_page(
@@ -20,7 +20,7 @@ register_page(
     title=get_title,
     description=get_description,
     order=50,
-    name=name,
+    name=NAME,
 )
 
 layout = html.Div(
@@ -39,29 +39,42 @@ def departement_marches(url):
     departement = url.split("/")[-1]
 
     def make_link_list(org_type) -> list:
-        link_list = []
+        table = (
-        if org_type == "acheteur":
+            "acheteurs_departement"
-            df = df_acheteurs_departement
+            if org_type == "acheteur"
-        elif org_type == "titulaire":
+            else "titulaires_departement"
-            df = df_titulaires_departement
+            if org_type == "titulaire"
-        else:
+            else None
+        )
+        if table is None:
             raise ValueError
+        col_prefix = org_type
+        rows = (
+            get_cursor()
+            .execute(
+                f"SELECT {col_prefix}_id, {col_prefix}_nom "
+                f"FROM {table} "
+                f"WHERE {col_prefix}_departement_code = ? "
+                f"ORDER BY {col_prefix}_nom",
+                [departement],
+            )
+            .fetchall()
+        )
 
-        df = df.filter(pl.col(f"{org_type}_departement_code") == departement)
+        link_list = []
-
+        for org_id, org_nom in rows:
-        for row in df.iter_rows(named=True):
             li = html.Li(
                 [
                     dcc.Link(
-                        row[f"{org_type}_nom"],
+                        org_nom,
-                        href=url + f"/{org_type}/{row[f'{org_type}_id']}",
+                        href=url + f"/{org_type}/{org_id}",
-                        title=f"Marchés publics de {row[f'{org_type}_nom']}",
+                        title=f"Marchés publics de {org_nom}",
                     ),
                     " ",
                     dcc.Link(
                         "(page dédiée)",
-                        href=f"/{org_type}s/{row[f'{org_type}_id']}",
+                        href=f"/{org_type}s/{org_id}",
-                        title=f"Page dédiée aux marchés publics de {row[f'{org_type}_nom']}",
+                        title=f"Page dédiée aux marchés publics de {org_nom}",
                     ),
                 ]
             )

src/pages/arbre/departements.py

@@ -1,8 +1,8 @@
 from dash import dcc, html, register_page
 
-from src.utils import departements
+from src.utils.data import DEPARTEMENTS
 
-name = "Départements"
+NAME = "Départements"
 
 register_page(
     __name__,
@@ -18,7 +18,7 @@ layout = html.Div(
         html.Ul(
             [
                 html.Li(dcc.Link(d["departement"], href=f"/departements/{k}"))
-                for k, d in departements.items()
+                for k, d in DEPARTEMENTS.items()
             ]
         ),
     ]

src/pages/arbre/liste_marches_org.py

@@ -1,22 +1,18 @@
 import polars as pl
 from dash import Input, Output, callback, dcc, html, register_page
 
-from src.utils import (
+from src.db import get_cursor
-    df_acheteurs,
+from src.utils.data import DF_ACHETEURS, DF_TITULAIRES
-    df_acheteurs_marches,
-    df_titulaires,
-    df_titulaires_marches,
-)
 
-name = "Liste des marchés publics"
+NAME = "Liste des marchés publics"
 
 
 def make_org_nom_verbe(org_type, org_id) -> tuple:
     if org_type == "titulaire":
-        df = df_titulaires
+        df = DF_TITULAIRES
         verbe = "remportés"
     elif org_type == "acheteur":
-        df = df_acheteurs
+        df = DF_ACHETEURS
         verbe = "attribués"
     else:
         raise ValueError
@@ -48,7 +44,7 @@ register_page(
     title=get_title,
     description=get_description,
     order=40,
-    name=name,
+    name=NAME,
 )
 
 layout = html.Div(
@@ -68,28 +64,34 @@ def liste_marches(url):
     org_id = url.split("/")[-1]
 
     def make_link_list() -> list:
-        link_list = []
+        table = (
-        if org_type == "acheteur":
+            "acheteurs_marches"
-            df = df_acheteurs_marches
+            if org_type == "acheteur"
-        elif org_type == "titulaire":
+            else "titulaires_marches"
-            df = df_titulaires_marches
+            if org_type == "titulaire"
-        else:
+            else None
+        )
+        if table is None:
             raise ValueError
+        rows = (
+            get_cursor()
+            .execute(
+                f"SELECT uid, objet FROM {table} WHERE {org_type}_id = ?",
+                [org_id],
+            )
+            .fetchall()
+        )
 
-        df = df.filter(pl.col(f"{org_type}_id") == org_id)
+        return [
-
+            html.Li(
-        for row in df.iter_rows(named=True):
-            li = html.Li(
-                [
                 dcc.Link(
-                        row["objet"],
+                    objet,
-                        href=f"/marches/{row['uid']}",
+                    href=f"/marches/{uid}",
-                        title=f"Marchés public attribué : {row['objet']}",
+                    title=f"Marchés public attribué : {objet}",
                 )
+            )
+            for uid, objet in rows
         ]
-            )
-            link_list.append(li)
-        return link_list
 
     nom, verbe = make_org_nom_verbe(org_type, org_id)
 

src/pages/marche.py

@@ -2,18 +2,13 @@ import json
 from datetime import datetime
 
 import dash_bootstrap_components as dbc
-import polars as pl
 from dash import Input, Output, callback, dcc, html, register_page
 from polars import selectors as cs
 
-from src.utils import (
+from src.db import query_marches
-    data_schema,
+from src.utils.data import DATA_SCHEMA
-    df,
+from src.utils.seo import META_CONTENT, make_org_jsonld
-    format_values,
+from src.utils.table import format_values, unformat_montant
-    make_org_jsonld,
-    meta_content,
-    unformat_montant,
-)
 
 
 def get_title(uid: str = None) -> str:
@@ -26,7 +21,7 @@ register_page(
     title=get_title,
     name="Marché",
     description="Consultez les détails de ce marché public : montant, acheteur, titulaires, modifications, etc.",
-    image_url=meta_content["image_url"],
+    image_url=META_CONTENT["image_url"],
     order=7,
 )
 
@@ -88,19 +83,17 @@ layout = [
 def get_marche_data(url) -> tuple[dict, list]:
     marche_uid = url.split("/")[-1]
 
-    # Récupération des données du marché à partir du df global
+    # Filtre SQL côté DuckDB, puis Polars pour le post-traitement
+    dff_marche = query_marches("uid = ?", (marche_uid,))
+    if dff_marche.height == 0:
+        return {}, []
 
-    lff = df.lazy()
+    lff = dff_marche.lazy()
-    lff = lff.filter(pl.col("uid") == pl.lit(marche_uid))
-
-    # Données des titulaires du marché
     dff_titulaires = lff.select(cs.starts_with("titulaire")).collect(engine="streaming")
+    dff_marche_unique = lff.unique("uid").collect(engine="streaming")
+    dff_marche_unique = format_values(dff_marche_unique)
 
-    # Données du marché
+    return dff_marche_unique.to_dicts()[0], dff_titulaires.to_dicts()
-    dff_marche = lff.unique("uid").collect(engine="streaming")
-    dff_marche = format_values(dff_marche)
-
-    return dff_marche.to_dicts()[0], dff_titulaires.to_dicts()
 
 
 @callback(
@@ -113,7 +106,7 @@ def get_marche_data(url) -> tuple[dict, list]:
 )
 def update_marche_info(marche, titulaires):
     def make_parameter(col, bold=True):
-        column_object = data_schema.get(col)
+        column_object = DATA_SCHEMA.get(col)
         column_name = column_object.get("title") if column_object else col
 
         if marche[col]:

src/pages/observatoire.py

@@ -16,6 +16,8 @@ from dash import (
     register_page,
 )
 
+from src.cache import cache
+from src.db import query_marches, schema
 from src.figures import (
     DataTable,
     get_barchart_sources,
@@ -28,50 +30,47 @@ from src.figures import (
     make_column_picker,
     make_donut,
 )
-from src.utils import (
+from src.utils import logger
-    columns,
+from src.utils.data import (
-    departements,
+    DEPARTEMENTS,
-    df,
+    DF_ACHETEURS,
-    df_acheteurs,
+    DF_TITULAIRES,
-    df_titulaires,
-    get_default_hidden_columns,
-    get_enum_values_as_dict,
-    logger,
-    meta_content,
     prepare_dashboard_data,
-    prepare_table_data,
 )
+from src.utils.frontend import get_enum_values_as_dict
+from src.utils.seo import META_CONTENT
+from src.utils.table import COLUMNS, get_default_hidden_columns, prepare_table_data
 
-name = "Observatoire"
+NAME = "Observatoire"
 
 register_page(
     __name__,
     path="/observatoire",
     title="Observatoire | decp.info",
-    name=name,
+    name=NAME,
     description="Visualisez l'état de la publication des données essentielles des marchés publics en France.",
-    image_url=meta_content["image_url"],
+    image_url=META_CONTENT["image_url"],
     order=3,
 )
-options_years = []
+OPTIONS_YEARS = []
 for year in reversed(range(2017, datetime.now().year + 1)):
     option_year = {
         "label": str(year),
         "value": year,
     }
-    options_years.append(option_year)
+    OPTIONS_YEARS.append(option_year)
 
-options_departements = []
+OPTIONS_DEPARTEMENTS = []
-for code in departements.keys():
+for code in DEPARTEMENTS.keys():
     departement = {
-        "label": f"{departements[code]['departement']} ({code})",
+        "label": f"{DEPARTEMENTS[code]['departement']} ({code})",
         "value": code,
     }
-    options_departements.append(departement)
+    OPTIONS_DEPARTEMENTS.append(departement)
 
 OBSERVATOIRE_COLUMNS = [
     col
-    for col in df.columns
+    for col in schema.names()
     if col.startswith("acheteur")
     or col.startswith("titulaire")
     or col
@@ -123,7 +122,7 @@ Alors, on fait comment ?
     html.Div(
         className="container-fluid",
         children=[
-            html.H2(children=[name], id="page_title"),
+            html.H2(children=[NAME], id="page_title"),
             dcc.Loading(
                 overlay_style={"visibility": "visible", "filter": "blur(2px)"},
                 id="loading-statistques",
@@ -141,7 +140,7 @@ Alors, on fait comment ?
                                         dbc.Col(
                                             dcc.Dropdown(
                                                 id="dashboard_year",
-                                                options=options_years,
+                                                options=OPTIONS_YEARS,
                                                 placeholder="12 derniers mois",
                                                 persistence=True,
                                                 persistence_type="local",
@@ -181,7 +180,7 @@ Alors, on fait comment ?
                                                 searchable=True,
                                                 multi=True,
                                                 placeholder="Département",
-                                                options=options_departements,
+                                                options=OPTIONS_DEPARTEMENTS,
                                                 persistence=True,
                                                 persistence_type="local",
                                             ),
@@ -220,7 +219,7 @@ Alors, on fait comment ?
                                                 searchable=True,
                                                 multi=True,
                                                 placeholder="Département",
-                                                options=options_departements,
+                                                options=OPTIONS_DEPARTEMENTS,
                                                 persistence=True,
                                                 persistence_type="local",
                                             ),
@@ -590,7 +589,6 @@ def sync_observatoire_share_url(*args):
     href = args[-1]
 
     if not href:
-        print("no update")
         return no_update, no_update
 
     base_url = href.split("?")[0]
@@ -607,7 +605,6 @@ def sync_observatoire_share_url(*args):
 
     query_string = urllib.parse.urlencode(params)
     full_url = f"{base_url}?{query_string}" if query_string else base_url
-    print("query", query_string)
 
     if params:
         copy_button = dcc.Clipboard(
@@ -650,39 +647,33 @@ def show_confirmation(n_clicks):
     return no_update
 
 
-@callback(
+def _normalize_filter_params(filter_params: dict) -> tuple:
-    Output("cards", "children"),
+    """Produce a deterministic, hashable key for caching."""
-    Output("observatoire-filters", "data"),
+    return tuple(
-    *[Input(fp[0], "value") for fp in FILTER_PARAMS],
+        sorted(
+            (k, tuple(v) if isinstance(v, list) else v)
+            for k, v in filter_params.items()
+        )
     )
-def udpate_dashboard_cards(*filter_values):
-    lff: pl.LazyFrame = df.lazy()
 
-    # Filtrage des données
-    filter_params = {}
-    for (input_id, url_key, is_multi, default), value in zip(
-        FILTER_PARAMS, filter_values
-    ):
-        filter_params[input_id] = value
 
-    print(filter_params)
+@cache.memoize()
+def _compute_dashboard_children(cache_key: tuple):
+    logger.debug("Cache miss — computing dashboard")
+    filter_params = {k: (list(v) if isinstance(v, tuple) else v) for k, v in cache_key}
 
+    lff: pl.LazyFrame = query_marches().lazy()
     lff = prepare_dashboard_data(lff=lff, **filter_params)
 
-    # Génération des métriques
     dff = lff.collect(engine="streaming")
 
-    logger.debug("Filter data: " + str(dff.height))
-
     df_per_uid = (
         dff.select("uid", "montant").group_by("uid").agg(pl.col("montant").first())
     )
     nb_marches = df_per_uid.height
 
     cards = []
-
     card_summary_table = get_dashboard_summary_table(dff, df_per_uid, nb_marches)
-
     cards.append(make_card(title="Résumé", paragraphs=card_summary_table))
 
     donut_acheteur_categorie, nb_acheteur_categories = make_donut(
@@ -741,10 +732,9 @@ def udpate_dashboard_cards(*filter_values):
     )
     cards.append(make_card(title="Top titulaires", fig=top_titulaires, lg=12, xl=8))
 
-    geographic_maps: list[dbc.Col] = get_geographic_maps(dff)
+    geographic_maps: list[dbc.Col] | None = get_geographic_maps(dff)
 
     other_cards = []
-
     sources_barchart = get_barchart_sources(lff, type_date="dateNotification")
     other_cards.append(
         make_card(
@@ -767,7 +757,25 @@ def udpate_dashboard_cards(*filter_values):
         )
     )
 
-    return dbc.Row(children=cards + geographic_maps + other_cards), filter_params
+    return cards + geographic_maps + other_cards
+
+
+@callback(
+    Output("cards", "children"),
+    Output("observatoire-filters", "data"),
+    *[Input(fp[0], "value") for fp in FILTER_PARAMS],
+)
+def update_dashboard_cards(*filter_values):
+    filter_params = {}
+    for (input_id, _url_key, _is_multi, _default), value in zip(
+        FILTER_PARAMS, filter_values
+    ):
+        filter_params[input_id] = value
+
+    cache_key = _normalize_filter_params(filter_params)
+    children = _compute_dashboard_children(cache_key)
+
+    return dbc.Row(children=children), filter_params
 
 
 @callback(
@@ -778,7 +786,7 @@ def udpate_dashboard_cards(*filter_values):
     prevent_initial_call=True,
 )
 def download_observatoire(_n_clicks, filter_params, hidden_columns):
-    lff = prepare_dashboard_data(lff=df.lazy(), **(filter_params or {}))
+    lff = prepare_dashboard_data(lff=query_marches().lazy(), **(filter_params or {}))
 
     if hidden_columns:
         lff = lff.drop(hidden_columns)
@@ -812,20 +820,20 @@ def add_organization_name_in_title(acheteur_id, titulaire_id):
         return match[nom_col].item(0) if match.height >= 1 else None
 
     if acheteur_id and len(acheteur_id) == 14:
-        if nom := lookup_nom(df_acheteurs, "acheteur_id", "acheteur_nom", acheteur_id):
+        if nom := lookup_nom(DF_ACHETEURS, "acheteur_id", "acheteur_nom", acheteur_id):
             return [
-                name,
+                NAME,
                 html.Small(nom, className="text-muted d-block fw-normal fs-5"),
             ]
     elif titulaire_id and len(titulaire_id) == 14:
         if nom := lookup_nom(
-            df_titulaires, "titulaire_id", "titulaire_nom", titulaire_id
+            DF_TITULAIRES, "titulaire_id", "titulaire_nom", titulaire_id
         ):
             return [
-                name,
+                NAME,
                 html.Small(nom, className="text-muted d-block fw-normal fs-5"),
             ]
-    return name
+    return NAME
 
 
 @callback(
@@ -869,7 +877,7 @@ def populate_preview_table(
     if not is_open:
         return (no_update,) * 9
 
-    lff = prepare_dashboard_data(lff=df.lazy(), **(filter_params or {}))
+    lff = prepare_dashboard_data(lff=query_marches().lazy(), **(filter_params or {}))
 
     return prepare_table_data(
         lff,
@@ -889,8 +897,8 @@ def populate_preview_table(
 )
 def update_hidden_columns_from_checkboxes(selected_columns):
     if selected_columns:
-        selected_columns = [columns[i] for i in selected_columns]
+        selected_columns = [COLUMNS[i] for i in selected_columns]
-        hidden_columns = [col for col in columns if col not in selected_columns]
+        hidden_columns = [col for col in COLUMNS if col not in selected_columns]
         return hidden_columns
     else:
         return []
@@ -918,7 +926,7 @@ def update_checkboxes_from_hidden_columns(hidden_cols, current_checkboxes):
     hidden_cols = hidden_cols or get_default_hidden_columns("tableau")
 
     # Show all columns that are NOT hidden
-    visible_cols = [columns.index(col) for col in columns if col not in hidden_cols]
+    visible_cols = [COLUMNS.index(col) for col in COLUMNS if col not in hidden_cols]
     return visible_cols
 
 

src/pages/recherche.py

@@ -2,23 +2,20 @@ import dash_bootstrap_components as dbc
 from dash import Input, Output, State, callback, dcc, html, register_page
 
 from src.figures import DataTable
-from src.utils import (
+from src.utils.data import DF_ACHETEURS, DF_TITULAIRES
-    df_acheteurs,
+from src.utils.search import search_org
-    df_titulaires,
+from src.utils.seo import META_CONTENT
-    meta_content,
+from src.utils.table import setup_table_columns
-    search_org,
-    setup_table_columns,
-)
 
-name = "Recherche"
+NAME = "Recherche"
 
 register_page(
     __name__,
     path="/",
     title="Recherche de marchés publics | decp.info",
-    name=name,
+    name=NAME,
     description="Explorez et analysez les données des marchés publics français avec cet outil libre et gratuit. Pour une commande publique accessible à toutes et tous.",
-    image_url=meta_content["image_url"],
+    image_url=META_CONTENT["image_url"],
     order=0,
 )
 
@@ -97,9 +94,9 @@ def update_search_results(n_submit, n_clicks, query):
 
         for org_type in ["acheteur", "titulaire"]:
             if org_type == "acheteur":
-                dff = df_acheteurs
+                dff = DF_ACHETEURS
             elif org_type == "titulaire":
-                dff = df_titulaires
+                dff = DF_TITULAIRES
             else:
                 raise ValueError(f"{org_type} is not supported")
 

src/pages/tableau.py

@@ -19,17 +19,16 @@ from dash import (
     register_page,
 )
 
+from src.db import query_marches, schema
 from src.figures import DataTable, make_column_picker
-from src.utils import (
+from src.utils import logger
-    columns,
+from src.utils.seo import META_CONTENT
-    df,
+from src.utils.table import (
+    COLUMNS,
     filter_table_data,
     get_default_hidden_columns,
     invert_columns,
-    logger,
-    meta_content,
     prepare_table_data,
-    schema,
     sort_table_data,
 )
 
@@ -38,18 +37,18 @@ update_date = datetime.fromtimestamp(update_date_timestamp).strftime("%d/%m/%Y")
 update_date_iso = datetime.fromtimestamp(update_date_timestamp).isoformat()
 
 
-name = "Tableau"
+NAME = "Tableau"
 register_page(
     __name__,
     path="/tableau",
     title="Tableau des marchés publics | decp.info",
-    name=name,
+    name=NAME,
     description="Consultez, filtrez et exportez les données essentielles de la commande publique sous forme de tableau.",
-    image_url=meta_content["image_url"],
+    image_url=META_CONTENT["image_url"],
     order=1,
 )
 
-datatable = html.Div(
+DATATABLE = html.Div(
     className="marches_table",
     children=DataTable(
         dtid="tableau_datatable",
@@ -61,7 +60,7 @@ datatable = html.Div(
         filter_action="custom",
         sort_action="custom",
         hidden_columns=[],
-        columns=[{"id": col, "name": col} for col in df.columns],
+        columns=[{"id": col, "name": col} for col in schema.names()],
     ),
 )
 
@@ -128,7 +127,7 @@ layout = [
         ],
     ),
     dcc.Markdown(
-        f"Ce tableau contient tous les marchés attribués en France. Il vous permet d'appliquer un filtre sur une ou plusieurs colonnes, et ainsi produire la liste de marchés dont vous avez besoin (exemples : [marchés de voirie < 40 k€ en 2025](/tableau?filtres=%7Bacheteur_id%7D+icontains+24350013900189+%26%26+%7BdateNotification%7D+icontains+2025%2A+%26%26+%7Bmontant%7D+i%3C+40000+%26%26+%7Bobjet%7D+icontains+voirie&colonnes=uid%2Cacheteur_id%2Cacheteur_nom%2Ctitulaire_id%2Ctitulaire_nom%2Cobjet%2Cmontant%2CdureeMois%2CdateNotification%2Cacheteur_departement_code%2CsourceDataset), [marchés > 500 k€ avec clause sociale attribués à des PME à plus de 100 km dans le Var](/tableau?filtres=%7Btitulaire_categorie%7D+icontains+PME+%26%26+%7Btitulaire_distance%7D+i%3E+100+%26%26+%7Bmontant%7D+i%3E+500000+%26%26+%7Bacheteur_departement_code%7D+icontains+83+%26%26+%7BconsiderationsSociales%7D+icontains+clause&colonnes=uid%2Cacheteur_id%2Cacheteur_nom%2Ctitulaire_id%2Ctitulaire_nom%2Cobjet%2Cmontant%2CdureeMois%2CdateNotification%2CconsiderationsSociales%2Ctitulaire_distance%2Cacheteur_departement_code%2Ctitulaire_categorie%2CsourceDataset)). Par défaut seules quelques colonnes sont affichées, mais vous pouvez en afficher jusqu'à {str(df.width)} en cliquant sur le bouton **Choisir les colonnes**. Cet outil est assez puissant, je vous recommande de lire le mode d'emploi pour en tirer pleinement partie.",
+        f"Ce tableau contient tous les marchés attribués en France. Il vous permet d'appliquer un filtre sur une ou plusieurs colonnes, et ainsi produire la liste de marchés dont vous avez besoin (exemples : [marchés de voirie < 40 k€ en 2025](/tableau?filtres=%7Bacheteur_id%7D+icontains+24350013900189+%26%26+%7BdateNotification%7D+icontains+2025%2A+%26%26+%7Bmontant%7D+i%3C+40000+%26%26+%7Bobjet%7D+icontains+voirie&colonnes=uid%2Cacheteur_id%2Cacheteur_nom%2Ctitulaire_id%2Ctitulaire_nom%2Cobjet%2Cmontant%2CdureeMois%2CdateNotification%2Cacheteur_departement_code%2CsourceDataset), [marchés > 500 k€ avec clause sociale attribués à des PME à plus de 100 km dans le Var](/tableau?filtres=%7Btitulaire_categorie%7D+icontains+PME+%26%26+%7Btitulaire_distance%7D+i%3E+100+%26%26+%7Bmontant%7D+i%3E+500000+%26%26+%7Bacheteur_departement_code%7D+icontains+83+%26%26+%7BconsiderationsSociales%7D+icontains+clause&colonnes=uid%2Cacheteur_id%2Cacheteur_nom%2Ctitulaire_id%2Ctitulaire_nom%2Cobjet%2Cmontant%2CdureeMois%2CdateNotification%2CconsiderationsSociales%2Ctitulaire_distance%2Cacheteur_departement_code%2Ctitulaire_categorie%2CsourceDataset)). Par défaut seules quelques colonnes sont affichées, mais vous pouvez en afficher jusqu'à {len(schema.names())} en cliquant sur le bouton **Choisir les colonnes**. Cet outil est assez puissant, je vous recommande de lire le mode d'emploi pour en tirer pleinement partie.",
         style={"maxWidth": "1000px"},
     ),
     html.Div(
@@ -188,7 +187,7 @@ layout = [
 
             ##### Afficher plus de colonnes
 
-            Par défaut, un nombre réduit de colonnes est affiché pour ne pas surcharger la page. Mais vous avez le choix parmi {str(df.width)} colonnes, ce serait dommage de vous limiter !
+            Par défaut, un nombre réduit de colonnes est affiché pour ne pas surcharger la page. Mais vous avez le choix parmi {len(schema.names())} colonnes, ce serait dommage de vous limiter !
 
             Pour afficher plus de colonnes, cliquez sur le bouton **Choisir les colonnes** et cochez les colonnes pour les afficher.
 
@@ -273,7 +272,7 @@ layout = [
                 scrollable=True,
                 size="xl",
             ),
-            datatable,
+            DATATABLE,
         ],
     ),
 ]
@@ -316,7 +315,7 @@ def update_table(href, page_current, page_size, filter_query, sort_by, data_time
     prevent_initial_call=True,
 )
 def download_data(n_clicks, filter_query, sort_by, hidden_columns: list = None):
-    lff: pl.LazyFrame = df.lazy()  # start from the original data
+    lff: pl.LazyFrame = query_marches().lazy()
 
     # Les colonnes masquées sont supprimées
     if hidden_columns:
@@ -481,8 +480,8 @@ def toggle_tableau_help(click_open, click_close, is_open):
 )
 def update_hidden_columns_from_checkboxes(selected_columns):
     if selected_columns:
-        selected_columns = [columns[i] for i in selected_columns]
+        selected_columns = [COLUMNS[i] for i in selected_columns]
-        hidden_columns = [col for col in columns if col not in selected_columns]
+        hidden_columns = [col for col in COLUMNS if col not in selected_columns]
         return hidden_columns
     else:
         return []
@@ -496,6 +495,8 @@ def update_hidden_columns_from_checkboxes(selected_columns):
     ),
 )
 def store_hidden_columns(hidden_columns):
+    if hidden_columns is None:
+        hidden_columns = get_default_hidden_columns("tableau")
     return hidden_columns
 
 
@@ -508,7 +509,7 @@ def update_checkboxes_from_hidden_columns(hidden_cols, current_checkboxes):
     hidden_cols = hidden_cols or get_default_hidden_columns("tableau")
 
     # Show all columns that are NOT hidden
-    visible_cols = [columns.index(col) for col in columns if col not in hidden_cols]
+    visible_cols = [COLUMNS.index(col) for col in COLUMNS if col not in hidden_cols]
     return visible_cols
 
 

src/pages/titulaire.py

@@ -15,6 +15,7 @@ from dash import (
     register_page,
 )
 
+from src.db import query_marches, schema
 from src.figures import (
     DataTable,
     get_distance_histogram,
@@ -22,24 +23,21 @@ from src.figures import (
     make_column_picker,
     point_on_map,
 )
-from src.utils import (
+from src.utils.data import DF_TITULAIRES, get_annuaire_data, get_departement_region
-    columns,
+from src.utils.frontend import get_button_properties
-    df,
+from src.utils.seo import META_CONTENT
-    df_titulaires,
+from src.utils.table import (
+    COLUMNS,
     filter_table_data,
     format_number,
-    get_annuaire_data,
-    get_button_properties,
     get_default_hidden_columns,
-    get_departement_region,
-    meta_content,
     prepare_table_data,
     sort_table_data,
 )
 
 
 def get_title(titulaire_id: str = None) -> str:
-    titulaire_nom = df_titulaires.filter(pl.col("titulaire_id") == titulaire_id).select(
+    titulaire_nom = DF_TITULAIRES.filter(pl.col("titulaire_id") == titulaire_id).select(
         "titulaire_nom"
     )
     if titulaire_nom.height > 0:
@@ -53,11 +51,11 @@ register_page(
     title=get_title,
     name="Titulaire",
     description="Consultez les marchés publics remportés par ce titulaire.",
-    image_url=meta_content["image_url"],
+    image_url=META_CONTENT["image_url"],
     order=5,
 )
 
-datatable = html.Div(
+DATATABLE = html.Div(
     className="marches_table",
     children=DataTable(
         dtid="titulaire_datatable",
@@ -69,7 +67,7 @@ datatable = html.Div(
         sort_action="custom",
         page_size=10,
         hidden_columns=[],
-        columns=[{"id": col, "name": col} for col in df.columns],
+        columns=[{"id": col, "name": col} for col in schema.names()],
     ),
 )
 
@@ -239,7 +237,7 @@ layout = [
                         scrollable=True,
                         size="xl",
                     ),
-                    datatable,
+                    DATATABLE,
                 ],
             ),
         ],
@@ -339,21 +337,18 @@ def update_titulaire_stats(data):
 )
 def get_titulaire_marches_data(url, titulaire_year: str) -> tuple:
     titulaire_siret = url.split("/")[-1]
-    lff = df.lazy()
+    lff = query_marches(
-    lff = lff.filter(
+        "titulaire_id = ? AND titulaire_typeIdentifiant = 'SIRET'",
-        (pl.col("titulaire_id") == titulaire_siret)
+        (titulaire_siret,),
-        & (pl.col("titulaire_typeIdentifiant") == "SIRET")
+    ).lazy()
-    )
     if titulaire_year and titulaire_year != "Toutes les années":
         lff = lff.filter(
             pl.col("dateNotification").cast(pl.String).str.starts_with(titulaire_year)
         )
     lff = lff.sort(["dateNotification", "uid"], descending=True, nulls_last=True)
     lff = lff.fill_null("")
-
     dff: pl.DataFrame = lff.collect(engine="streaming")
     download_disabled, download_text, download_title = get_button_properties(dff.height)
-
     data = dff.to_dicts()
     return data, download_disabled, download_text, download_title
 
@@ -479,22 +474,23 @@ clientside_callback(
 )
 def update_hidden_columns_from_checkboxes(selected_columns):
     if selected_columns:
-        selected_columns = [columns[i] for i in selected_columns]
+        selected_columns = [COLUMNS[i] for i in selected_columns]
-        hidden_columns = [col for col in columns if col not in selected_columns]
+        hidden_columns = [col for col in COLUMNS if col not in selected_columns]
         return hidden_columns
     else:
         return []
 
 
 @callback(
-    Output("titulaire_datatable", "hidden_columns", allow_duplicate=True),
+    Output("titulaire_datatable", "hidden_columns"),
     Input(
         "titulaire-hidden-columns",
         "data",
     ),
-    prevent_initial_call=True,
 )
 def store_hidden_columns(hidden_columns):
+    if hidden_columns is None:
+        hidden_columns = get_default_hidden_columns("titulaire")
     return hidden_columns
 
 
@@ -507,7 +503,7 @@ def update_checkboxes_from_hidden_columns(hidden_cols, current_checkboxes):
     hidden_cols = hidden_cols or get_default_hidden_columns("titulaire")
 
     # Show all columns that are NOT hidden
-    visible_cols = [columns.index(col) for col in columns if col not in hidden_cols]
+    visible_cols = [COLUMNS.index(col) for col in COLUMNS if col not in hidden_cols]
     return visible_cols
 
 

src/utils.py

@@ -1,916 +0,0 @@
-import json
-import logging
-import os
-import uuid
-from collections import OrderedDict
-from datetime import datetime, timedelta
-from time import localtime, sleep
-
-import polars as pl
-import polars.selectors as cs
-from dash import no_update
-from httpx import HTTPError, get, post
-from polars.exceptions import ComputeError
-from unidecode import unidecode
-
-logging.basicConfig(
-    format="%(asctime)s %(levelname)-8s %(message)s",
-    level=logging.INFO,
-    datefmt="%Y-%m-%d %H:%M:%S",
-)
-logger = logging.getLogger("decp.info")
-development = os.getenv("DEVELOPMENT", "False").lower() == "true"
-if development:
-    logger.setLevel(logging.DEBUG)
-
-logging.getLogger("httpx").setLevel("WARNING")
-
-
-def split_filter_part(filter_part):
-    operators = [
-        ["s<", "<"],
-        ["s>", ">"],
-        ["i<", "<"],
-        ["i>", ">"],
-        ["icontains", "contains"],
-        # [" ", "contains"]
-    ]
-    logger.debug("filter part " + filter_part)
-    for operator_group in operators:
-        if operator_group[0] in filter_part:
-            name_part, value_part = filter_part.split(operator_group[0], 1)
-            name_part = name_part.strip()
-            value = value_part.strip()
-            name = name_part[name_part.find("{") + 1 : name_part.rfind("}")]
-            logger.debug("=> " + " ".join([name, operator_group[1], value]))
-
-            return name, operator_group[1], value
-
-    return [None] * 3
-
-
-def add_resource_link(dff: pl.DataFrame) -> pl.DataFrame:
-    dff = dff.with_columns(
-        (
-            '<a href="' + pl.col("sourceFile") + '">' + pl.col("sourceDataset") + "</a>"
-        ).alias("sourceDataset")
-    )
-    dff = dff.drop(["sourceFile"])
-    return dff
-
-
-def add_links(dff: pl.DataFrame):
-    for col in ["uid", "acheteur_nom", "titulaire_nom", "acheteur_id", "titulaire_id"]:
-        if col in dff.columns:
-            if col.startswith("titulaire_"):
-                detail_link = (
-                    '<a href = "/titulaires/'
-                    + pl.col("titulaire_id")
-                    + '">'
-                    + pl.col(col)
-                    + "</a>"
-                )
-                if col == "titulaire_nom":
-                    detail_link = (
-                        detail_link
-                        + ' <a href="/observatoire?titulaire_id='
-                        + pl.col("titulaire_id")
-                        + '" title="Voir dans l\'observatoire">📊</a>'
-                    )
-                dff = dff.with_columns(
-                    pl.when(
-                        pl.Expr.or_(
-                            pl.col("titulaire_typeIdentifiant").is_null(),
-                            pl.col("titulaire_typeIdentifiant") == "SIRET",
-                        )
-                    )
-                    .then(detail_link)
-                    .otherwise(pl.col(col))
-                    .alias(col)
-                )
-            if col.startswith("acheteur_"):
-                detail_link = (
-                    '<a href = "/acheteurs/'
-                    + pl.col("acheteur_id")
-                    + '">'
-                    + pl.col(col)
-                    + "</a>"
-                )
-                if col == "acheteur_nom":
-                    detail_link = (
-                        detail_link
-                        + ' <a href="/observatoire?acheteur_id='
-                        + pl.col("acheteur_id")
-                        + '" title="Voir dans l\'observatoire">📊</a>'
-                    )
-                dff = dff.with_columns(detail_link.alias(col))
-            if col == "uid":
-                dff = dff.with_columns(
-                    (
-                        '<a href = "/marches/'
-                        + pl.col("uid")
-                        + '">'
-                        + pl.col("uid")
-                        + "</a>"
-                    ).alias("uid")
-                )
-    return dff
-
-
-def add_links_in_dict(data: list[dict], org_type: str) -> list:
-    new_data = []
-    for marche in data:
-        org_id = marche[org_type + "_id"]
-        marche[org_type + "_nom"] = (
-            f'<a href="/{org_type}s/{org_id}">{marche[org_type + "_nom"]}</a>'
-        )
-        if marche.get("uid"):
-            marche["id"] = f'<a href="/marches/{marche["uid"]}">{marche["id"]}</a>'
-            marche["uid"] = f'<a href="/marches/{marche["uid"]}">{marche["uid"]}</a>'
-        new_data.append(marche)
-    return new_data
-
-
-def booleans_to_strings(lff: pl.LazyFrame) -> pl.LazyFrame:
-    """
-    Convert all boolean columns to string type.
-    """
-    lff = lff.with_columns(
-        pl.col(cs.Boolean)
-        .cast(pl.String)
-        .str.replace("true", "oui")
-        .str.replace("false", "non")
-    )
-    return lff
-
-
-def numbers_to_strings(lff: pl.LazyFrame) -> pl.LazyFrame:
-    """
-    Convert all numeric columns to string type.
-    """
-    lff = lff.with_columns(pl.col(pl.Float64, pl.Int16).cast(pl.String).fill_null(""))
-    return lff
-
-
-def dates_to_strings(lff: pl.LazyFrame, column: str) -> pl.LazyFrame:
-    """
-    Convert a date column to string type.
-    """
-    lff = lff.with_columns(pl.col(column).cast(pl.String).fill_null(""))
-    return lff
-
-
-def format_number(number) -> str:
-    number = "{:,}".format(number).replace(",", " ")
-    return number
-
-
-def unformat_montant(number: str) -> float:
-    number = number.replace(" €", "")
-    number = number.replace(" €", "").replace(" ", "")
-    number = number.replace(",", ".")
-    number = number.strip()
-    return float(number)
-
-
-def format_values(dff: pl.DataFrame) -> pl.DataFrame:
-    def format_montant(expr, scale=None):
-        # https://stackoverflow.com/a/78636786
-        expr = expr.cast(pl.String)
-        expr = expr.str.splitn(".", 2)
-
-        num = expr.struct[0]
-        frac = expr.struct[1]
-
-        # Ajout des espaces
-        num = (
-            num.str.reverse()
-            .str.replace_all(r"\d{3}", "$0 ")
-            .str.reverse()
-            .str.replace(r"^ ", "")
-        )
-
-        frac: pl.Expr = (
-            pl.when(frac.is_not_null() & ~frac.is_in(["0"]))
-            .then("," + frac.str.head(2))
-            .otherwise(pl.lit(""))
-        )
-
-        montant: pl.Expr = (
-            pl.when((num + frac) == pl.lit(""))
-            .then(pl.lit(""))
-            .otherwise(num + frac + pl.lit(" €"))
-        )
-
-        return montant
-
-    def format_distance(expr):
-        expr = expr.cast(pl.String)
-        return pl.concat_str(expr, pl.lit(" km"))
-
-    if "montant" in dff.columns:
-        dff = dff.with_columns(pl.col("montant").pipe(format_montant).alias("montant"))
-    if "titulaire_distance" in dff.columns:
-        dff = dff.with_columns(
-            pl.col("titulaire_distance")
-            .pipe(format_distance)
-            .alias("titulaire_distance")
-        )
-
-    return dff
-
-
-def get_annuaire_data(siret: str) -> dict:
-    url = f"https://recherche-entreprises.api.gouv.fr/search?q={siret}"
-    try:
-        response = get(url).raise_for_status()
-        response = response.json()["results"][0]
-    except (HTTPError, IndexError):
-        response = None
-        logger.warning("Could not fetch data from recherche-entreprises.api.")
-    return response
-
-
-def get_decp_data() -> pl.DataFrame:
-    # Chargement du fichier parquet
-    # Le fichier est chargé en mémoire, ce qui est plus rapide qu'une base de données pour le moment.
-    # On utilise polars pour la rapidité et la facilité de manipulation des données.
-
-    try:
-        logger.info(
-            f"Lecture du fichier parquet ({os.getenv('DATA_FILE_PARQUET_PATH')})..."
-        )
-        lff: pl.LazyFrame = pl.scan_parquet(os.getenv("DATA_FILE_PARQUET_PATH"))
-    except ComputeError:
-        # Le fichier est probablement en cours de mise à jour
-        logger.info("Échec, nouvelle tentative dans 10s...")
-        sleep(10)
-        lff: pl.LazyFrame = pl.scan_parquet(os.getenv("DATA_FILE_PARQUET_PATH"))
-
-    # Tri des marchés par date de notification
-    lff = lff.sort(by=["dateNotification", "uid"], descending=True, nulls_last=True)
-
-    # Uniquement les données actuelles, pas les anciennes versions de marchés
-    lff = lff.filter(pl.col("donneesActuelles")).drop("donneesActuelles")
-
-    # Convertir les colonnes booléennes en chaînes de caractères
-    lff = booleans_to_strings(lff)
-
-    # Mention pour les org dont on a pas le nom
-    for col in ["acheteur_nom", "titulaire_nom"]:
-        lff = lff.with_columns(
-            pl.when(pl.col(col).is_null())
-            .then(pl.lit("[Identifiant non reconnu dans la base INSEE]"))
-            .otherwise(pl.col(col))
-            .name.keep()
-        )
-
-    # Bizarrement je ne peux pas faire lff = lff.fill_null("") ici
-    # ça génère une erreur dans la page acheteur (acheteur_data.table) :
-    # AttributeError: partially initialized module 'pandas' has no attribute 'NaT' (most likely due to a circular import)
-
-    return lff.collect()
-
-
-def get_org_data(dff: pl.DataFrame, org_type: str) -> pl.DataFrame:
-    lff = dff.lazy()
-    lff = lff.select(
-        "uid",
-        cs.starts_with(org_type).exclude(
-            f"{org_type}_latitude", f"{org_type}_longitude"
-        ),
-    )
-    lff = lff.group_by(cs.starts_with(org_type)).len("Marchés")
-    return lff.collect()
-
-
-def get_statistics() -> dict:
-    return (
-        get(
-            "https://www.data.gouv.fr/api/1/datasets/r/0ccf4a75-f3aa-4b46-8b6a-18aeb63e36df",
-            follow_redirects=True,
-        )
-        .raise_for_status()
-        .json()
-    )
-
-
-def get_departements() -> dict:
-    with open("data/departements.json", "rb") as f:
-        data = json.load(f)
-        return data
-
-
-def get_departements_geojson() -> dict:
-    with open("./data/departements-1000m.geojson") as f:
-        geojson = json.load(f)
-
-    # Ajout de feature.id
-    for f in geojson["features"]:
-        f["id"] = f["properties"]["code"]
-
-    return geojson
-
-
-def get_departement_region(code_postal):
-    if code_postal > "97000":
-        code_departement = code_postal[:3]
-    else:
-        code_departement = code_postal[:2]
-    nom_departement = departements[code_departement]["departement"]
-    nom_region = departements[code_departement]["region"]
-    return code_departement, nom_departement, nom_region
-
-
-def filter_table_data(
-    lff: pl.LazyFrame, filter_query: str, filter_source: str
-) -> pl.LazyFrame:
-    _schema = lff.collect_schema()
-    track_search(filter_query, filter_source)
-    filtering_expressions = filter_query.split(" && ")
-    for filter_part in filtering_expressions:
-        col_name, operator, filter_value = split_filter_part(filter_part)
-        col_type = str(_schema[col_name])
-        # logger.debug("filter_value:", filter_value)
-        # logger.debug("filter_value_type:", type(filter_value))
-        # logger.debug("operator:", operator)
-        # logger.debug("col_type:", col_type)
-
-        lff = lff.filter(pl.col(col_name).is_not_null())
-
-        if col_type == "Date":
-            # Convertir la colonne date en chaînes de caractères
-            lff = dates_to_strings(lff, col_name)
-            col_type = "String"
-        if col_type == "String":
-            lff = lff.filter(pl.col(col_name) != pl.lit(""))
-
-        elif col_type.startswith("Int") or col_type.startswith("Float"):
-            try:
-                filter_value = int(filter_value)
-            except ValueError:
-                logger.error(f"Invalid numeric filter value: {filter_value}")
-                continue
-
-        if operator in ("contains", "<", "<=", ">", ">="):
-            if operator == "<":
-                lff = lff.filter(pl.col(col_name) < filter_value)
-            elif operator == ">":
-                lff = lff.filter(pl.col(col_name) > filter_value)
-            elif operator == ">=":
-                lff = lff.filter(pl.col(col_name) >= filter_value)
-            elif operator == "<=":
-                lff = lff.filter(pl.col(col_name) <= filter_value)
-            elif operator == "contains":
-                if col_type in ["String", "Date"]:
-                    filter_value = filter_value.strip('"')
-                    if filter_value.endswith("*"):
-                        lff = lff.filter(
-                            pl.col(col_name).str.starts_with(filter_value[:-1])
-                        )
-                    elif filter_value.startswith("*"):
-                        lff = lff.filter(
-                            pl.col(col_name).str.ends_with(filter_value[1:])
-                        )
-                    else:
-                        lff = lff.filter(
-                            pl.col(col_name).str.contains("(?i)" + filter_value)
-                        )
-                elif col_type.startswith("Int") or col_type.startswith("Float"):
-                    lff = lff.filter(pl.col(col_name) == filter_value)
-                else:
-                    logger.error(f"Invalid column type: {col_type}")
-        else:
-            logger.error(f"Invalid operator: {operator}")
-
-        # elif operator == 'datestartswith':
-        # lff = lff.filter(pl.col(col_name).str.startswith(filter_value)")
-
-    return lff
-
-
-def sort_table_data(lff: pl.LazyFrame, sort_by: list) -> pl.LazyFrame:
-    lff = lff.sort(
-        [col["column_id"] for col in sort_by],
-        descending=[col["direction"] == "desc" for col in sort_by],
-        nulls_last=True,
-    )
-    logger.debug(sort_by)
-    return lff
-
-
-def setup_table_columns(
-    dff, hideable: bool = True, exclude: list = None, new_columns: list = None
-) -> tuple:
-    # Liste finale de colonnes
-    markdown_exceptions = ["montant", "titulaire_distance", "distance", "dureeMois"]
-    columns = []
-    tooltip = {}
-    for column_id in dff.columns:
-        if exclude and column_id in exclude:
-            continue
-        column_object = data_schema.get(column_id)
-        if column_object:
-            column_name = column_object.get("title")
-        else:
-            # Si le champ est un champ créé par erreur lors d'une jointure, on le skip
-            if column_id.endswith("_left") or column_id.endswith("_right"):
-                logger.warning(f"Champ innatendu : {column_id}")
-                continue
-            column_name = column_id
-            column_object = {"title": column_name, "description": ""}
-
-        presentation = "input" if column_id in markdown_exceptions else "markdown"
-
-        column = {
-            "name": column_name,
-            "id": column_id,
-            "presentation": presentation,
-            "type": "text",
-            "format": {"nully": "N/A"},
-            "hideable": hideable,
-        }
-        columns.append(column)
-        if column_object:
-            tooltip[column_id] = {
-                "value": f"""**{column_object.get("title")}** ({column_id})
-
-    """
-                + column_object.get("description", ""),
-                "type": "markdown",
-            }
-    return columns, tooltip
-
-
-def get_default_hidden_columns(page):
-    if page == "acheteur":
-        displayed_columns = [
-            "uid",
-            "objet",
-            "dateNotification",
-            "titulaire_id",
-            "titulaire_typeIdentifiant",
-            "titulaire_nom",
-            "titulaire_distance",
-            "montant",
-            "codeCPV",
-            "dureeRestanteMois",
-        ]
-    elif page == "titulaire":
-        displayed_columns = [
-            "uid",
-            "objet",
-            "dateNotification",
-            "acheteur_id",
-            "acheteur_nom",
-            "titulaire_distance",
-            "montant",
-            "codeCPV",
-            "dureeRestanteMois",
-        ]
-    elif page == "tableau":
-        displayed_columns = os.getenv("DISPLAYED_COLUMNS")
-    else:
-        displayed_columns = os.getenv("DISPLAYED_COLUMNS")
-        logger.warning(f"Invalid page: {page}")
-
-    hidden_columns = []
-
-    for col in schema.names():
-        if col in displayed_columns:
-            continue
-        else:
-            hidden_columns.append(col)
-    return hidden_columns
-
-
-def get_data_schema() -> dict:
-    # Récupération du schéma des données tabulaires
-    path = os.getenv("DATA_SCHEMA_PATH")
-    if path.startswith("http"):
-        original_schema: dict = get(
-            os.getenv("DATA_SCHEMA_PATH"), follow_redirects=True
-        ).json()
-    elif os.path.exists(path):
-        with open(path) as f:
-            original_schema: dict = json.load(f)
-    else:
-        raise Exception(f"Chemin vers le schéma invalide: {path}")
-
-    new_schema = OrderedDict()
-
-    for col in original_schema["fields"]:
-        new_schema[col["name"]] = col
-
-    return new_schema
-
-
-def track_search(query, category):
-    if len(query) >= 4 and not development and os.getenv("MATOMO_DOMAIN"):
-        url = "https://decp.info"
-        params = {
-            "idsite": os.getenv("MATOMO_ID_SITE"),
-            "url": url,
-            "rec": "1",
-            "action_name": "search" if category == "home_page_search" else "filter",
-            "search_cat": category,
-            "rand": uuid.uuid4().hex,
-            "apiv": "1",
-            "h": localtime().tm_hour,
-            "m": localtime().tm_min,
-            "s": localtime().tm_sec,
-            "search": query,
-            "token_auth": os.getenv("MATOMO_TOKEN"),
-        }
-        post(
-            url=f"https://{os.getenv('MATOMO_DOMAIN')}/matomo.php",
-            params=params,
-        ).raise_for_status()
-
-
-def search_org(dff: pl.DataFrame, query: str, org_type: str) -> pl.DataFrame:
-    """
-    Search in either 'acheteur' or 'titulaire' DataFrame.
-
-    :param dff: Polars DataFrame with acheteur or titulaire columns
-    :param query: User search string
-    :param org_type: 'acheteur' or 'titulaire'
-    :return: Filtered DataFrame with 'matches' column
-    """
-    if not query.strip():
-        return dff.select(pl.lit(False).alias("matches"))
-
-    # Enregistrement des recherche dans Matomo
-    track_search(query, "home_page_search")
-
-    # Normalize query
-    normalized_query = unidecode(query.strip()).upper()
-    tokens = [" " + t.strip() for t in normalized_query.split() if t.strip()]
-
-    # Define columns based on entity type
-    cols = [
-        f"{org_type}_id",
-        f"{org_type}_nom",
-        f"{org_type}_departement_nom",
-        f"{org_type}_departement_code",
-        f"{org_type}_commune_nom",
-    ]
-
-    # Concatenate all fields into one string per row
-    org_str = pl.concat_str(pl.lit(" "), pl.col(cols), separator=" ").str.replace(
-        "-", " "
-    )
-
-    # For each token, create a boolean column: True if token is found
-    token_matches = []
-    for token in tokens:
-        token_match = org_str.str.contains(token).alias(f"token_{token}")
-        token_matches.append(token_match)
-
-    # Count how many tokens match per row
-    match_score = pl.sum_horizontal(token_matches).alias("match_score")
-
-    # For each token, create a boolean column: True if token is found
-    token_matches = []
-    for token in tokens:
-        token_match = org_str.str.contains(token).alias(f"token_{token}")
-        token_matches.append(token_match)
-
-    # Sélection des colonnes
-    if org_type == "acheteur":
-        dff = dff.select(cols + ["Marchés"])
-    if org_type == "titulaire":
-        dff = dff.select(cols + ["Marchés", "titulaire_typeIdentifiant"])
-
-    # Apply and filter
-    dff = (
-        dff.with_columns(token_matches + [match_score])
-        .filter(pl.col("match_score") == len(tokens))
-        .drop([f"token_{token}" for token in tokens])
-    )
-
-    # Format result
-    dff = add_links(dff)
-    dff = dff.with_columns(
-        pl.concat_str(
-            pl.col(f"{org_type}_departement_nom"),
-            pl.lit(" ("),
-            pl.col(f"{org_type}_departement_code"),
-            pl.lit(")"),
-        ).alias("Département")
-    )
-
-    dff = dff.select(f"{org_type}_id", f"{org_type}_nom", "Département", "Marchés")
-    dff = dff.group_by(f"{org_type}_id", f"{org_type}_nom", "Département").sum()
-    dff = dff.sort("Marchés", descending=True)
-
-    return dff
-
-
-def prepare_table_data(
-    data, data_timestamp, filter_query, page_current, page_size, sort_by, source_table
-):
-    """
-    Fonction de préparation des données pour les datatables, afin de permettre une gestion fine des logiques,
-    notamment pour les filtres et les tris.
-    :param data
-    :param data_timestamp:
-    :param filter_query:
-    :param page_current:
-    :param page_size:
-    :param sort_by:
-    :param source_table:
-    :return:
-    """
-
-    if os.getenv("DEVELOPMENT").lower() == "true":
-        logger.debug(" + + + + + + + + + + + + + + + + + + ")
-
-    trigger_cleanup = no_update
-
-    # Récupération des données
-    if isinstance(data, list):
-        lff: pl.LazyFrame = pl.LazyFrame(data, strict=False, infer_schema_length=5000)
-    elif isinstance(data, pl.LazyFrame):
-        lff = data
-    else:
-        lff: pl.LazyFrame = df.lazy()  # start from the original data
-
-    # Application des filtres
-    if filter_query:
-        lff = filter_table_data(lff, filter_query, source_table)
-        trigger_cleanup = no_update if source_table == "tableau" else str(uuid.uuid4())
-
-    # Application des tris
-    if sort_by and len(sort_by) > 0:
-        lff = sort_table_data(lff, sort_by)
-
-    # Matérialisation des filtres
-    dff: pl.DataFrame = lff.collect()
-    height = dff.height
-
-    if height > 0:
-        nb_rows = f"{format_number(height)} lignes ({format_number(dff.select('uid').unique().height)} marchés)"
-    else:
-        nb_rows = "0 lignes (0 marchés)"
-
-    # Pagination des données
-    start_row = page_current * page_size
-    # end_row = (page_current + 1) * page_size
-    dff = dff.slice(start_row, page_size)
-
-    # Tout devient string
-    dff = dff.cast(pl.String)
-
-    # Remplace les strings null par "", mais pas les numeric null
-    dff = dff.fill_null("")
-
-    # Ajout des liens vers les pages de détails
-    dff = add_links(dff)
-
-    # Ajout des liens vers les fichiers Open Data
-    if "sourceFile" in dff.columns:
-        dff = add_resource_link(dff)
-
-    # Formatage des montants
-    if height > 0:
-        dff = format_values(dff)
-
-    # Récupération des colonnes et tooltip
-    table_columns, tooltip = setup_table_columns(dff)
-
-    dicts = dff.to_dicts()
-
-    # Propriétés du bouton de téléchargement
-    download_disabled, download_text, download_title = get_button_properties(height)
-
-    return (
-        dicts,
-        table_columns,
-        tooltip,
-        data_timestamp + 1,
-        nb_rows,
-        download_disabled,
-        download_text,
-        download_title,
-        trigger_cleanup,
-    )
-
-
-def prepare_dashboard_data(
-    lff: pl.LazyFrame,
-    dashboard_year,
-    dashboard_acheteur_id,
-    dashboard_acheteur_categorie,
-    dashboard_acheteur_departement_code,
-    dashboard_titulaire_id,
-    dashboard_titulaire_categorie,
-    dashboard_titulaire_departement_code,
-    dashboard_marche_type,
-    dashboard_marche_objet,
-    dashboard_marche_code_cpv,
-    dashboard_marche_considerations_sociales,
-    dashboard_marche_considerations_environnementales,
-    dashboard_marche_techniques,
-    dashboard_marche_innovant,
-    dashboard_marche_sous_traitance_declaree,
-    dashboard_montant_min=None,
-    dashboard_montant_max=None,
-) -> pl.LazyFrame:
-    if dashboard_year:
-        lff = lff.filter(pl.col("dateNotification").dt.year() == int(dashboard_year))
-    else:
-        lff = lff.filter(
-            pl.col("dateNotification") > (datetime.now() - timedelta(days=365))
-        )
-
-    if dashboard_acheteur_id:
-        lff = lff.filter(pl.col("acheteur_id").str.contains(dashboard_acheteur_id))
-    else:
-        if dashboard_acheteur_categorie:
-            lff = lff.filter(pl.col("acheteur_categorie") == dashboard_acheteur_categorie)
-        if dashboard_acheteur_departement_code:
-            lff = lff.filter(
-                pl.col("acheteur_departement_code").is_in(dashboard_acheteur_departement_code)
-            )
-
-    if dashboard_titulaire_id:
-        lff = lff.filter(pl.col("titulaire_id").str.contains(dashboard_titulaire_id))
-    else:
-        if dashboard_titulaire_categorie:
-            lff = lff.filter(pl.col("titulaire_categorie") == dashboard_titulaire_categorie)
-        if dashboard_titulaire_departement_code:
-            lff = lff.filter(
-                pl.col("titulaire_departement_code").is_in(dashboard_titulaire_departement_code)
-            )
-
-    if dashboard_marche_type:
-        lff = lff.filter(pl.col("type") == dashboard_marche_type)
-
-    if dashboard_marche_objet:
-        lff = lff.filter(pl.col("objet").str.contains(f"(?i){dashboard_marche_objet}"))
-
-    if dashboard_marche_code_cpv:
-        lff = lff.filter(pl.col("codeCPV").str.starts_with(dashboard_marche_code_cpv))
-
-    if dashboard_marche_innovant and dashboard_marche_innovant != "all":
-        lff = lff.filter(pl.col("marcheInnovant") == dashboard_marche_innovant)
-
-    if dashboard_marche_sous_traitance_declaree and dashboard_marche_sous_traitance_declaree != "all":
-        lff = lff.filter(pl.col("sousTraitanceDeclaree") == dashboard_marche_sous_traitance_declaree)
-
-    if dashboard_marche_techniques:
-        lff = lff.filter(
-            pl.col("techniques")
-            .str.split(", ")
-            .list.set_intersection(dashboard_marche_techniques)
-            .list.len()
-            > 0
-        )
-
-    if dashboard_marche_considerations_sociales:
-        lff = lff.filter(
-            pl.col("considerationsSociales")
-            .str.split(", ")
-            .list.set_intersection(dashboard_marche_considerations_sociales)
-            .list.len()
-            > 0
-        )
-
-    if dashboard_marche_considerations_environnementales:
-        lff = lff.filter(
-            pl.col("considerationsEnvironnementales")
-            .str.split(", ")
-            .list.set_intersection(dashboard_marche_considerations_environnementales)
-            .list.len()
-            > 0
-        )
-
-    if dashboard_montant_min is not None:
-        lff = lff.filter(pl.col("montant") >= dashboard_montant_min)
-
-    if dashboard_montant_max is not None:
-        lff = lff.filter(pl.col("montant") <= dashboard_montant_max)
-
-    return lff
-
-
-def get_button_properties(height):
-    if height > 65000:
-        download_disabled = True
-        download_text = "Téléchargement désactivé au-delà de 65 000 lignes"
-        download_title = " Ajoutez des filtres pour réduire le nombre de lignes, Excel ne supporte pas d'avoir plus de 65 000 URLs dans une même feuille de calcul."
-    elif height == 0:
-        download_disabled = True
-        download_text = "Pas de données à télécharger"
-        download_title = ""
-    else:
-        download_disabled = False
-        download_text = "Télécharger au format Excel"
-        download_title = "Télécharger les données telles qu'affichées au format Excel"
-    return download_disabled, download_text, download_title
-
-
-def get_enum_values_as_dict(column_name):
-    try:
-        options = {}
-        for value in data_schema[column_name]["enum"]:
-            options[value] = value
-        return options
-    except KeyError:
-        return {"not_found": "not found"}
-
-
-def invert_columns(columns):
-    """
-    Renvoie les colonnes du schéma non spécifiées en paramètre. Utile pour passer d'une colonnes masquées à une liste de colonnes affichées, et vice versa.
-
-    :param columns:
-    :return:
-    """
-    inverted_columns = []
-    for column in schema.names():
-        if column not in columns:
-            inverted_columns.append(column)
-    return inverted_columns
-
-
-def make_org_jsonld(org_id, org_type, org_name=None, type_org_id="SIRET") -> dict:
-    org_types = {"acheteur": "GovernmentOrganization", "titulaire": "Organization"}
-    address = None
-    if type_org_id.lower() == "siret" and len(org_id) == 14:
-        annuaire_data = get_annuaire_data(org_id)
-        annuaire_address = annuaire_data["matching_etablissements"][0]
-        code_postal = annuaire_address["code_postal"]
-        commune = annuaire_address["libelle_commune"]
-
-        address = (
-            {
-                "@type": "PostalAddress",
-                "streetAddress": annuaire_address.get("adresse", "")
-                .replace(code_postal, "")
-                .replace(commune, "")
-                .strip(),
-                "addressLocality": commune,
-                "postalCode": code_postal,
-                "addressCountry": "FR",
-            },
-        )
-
-    jsonld = {
-        "@type": org_types[org_type],
-        "name": org_name,
-        "url": f"https://decp.info/{org_type}s/{org_id}",
-        "sameAs": f"https://annuaire-entreprises.data.gouv.fr/etablissement/{org_id}",
-        "identifier": {
-            "@type": "PropertyValue",
-            "propertyID": type_org_id.lower(),
-            "value": org_id,
-        },
-    }
-
-    if address:
-        jsonld["address"] = address
-
-    return jsonld
-
-
-df: pl.DataFrame = get_decp_data()
-schema = df.collect_schema()
-
-df_acheteurs = get_org_data(df, "acheteur")
-df_titulaires = get_org_data(df, "titulaire")
-df_acheteurs_departement: pl.DataFrame = (
-    df_acheteurs.select(["acheteur_id", "acheteur_nom", "acheteur_departement_code"])
-    .unique()
-    .sort("acheteur_nom")
-)
-df_titulaires_departement: pl.DataFrame = (
-    df_titulaires.select(
-        ["titulaire_id", "titulaire_nom", "titulaire_departement_code"]
-    )
-    .unique()
-    .sort("titulaire_nom")
-)
-df_acheteurs_marches: pl.DataFrame = (
-    df.select("uid", "objet", "acheteur_id").unique().sort("acheteur_id")
-)
-df_titulaires_marches: pl.DataFrame = (
-    df.select("uid", "objet", "titulaire_id").unique().sort("titulaire_id")
-)
-
-departements = get_departements()
-departements_geojson = get_departements_geojson()
-domain_name = (
-    "test.decp.info" if os.getenv("DEVELOPMENT").lower() == "true" else "decp.info"
-)
-meta_content = {
-    "image_url": f"https://{domain_name}/assets/decp.info.png",
-    "title": "decp.info - exploration des marchés publics français",
-    "description": (
-        "Explorez et analysez les données des marchés publics français avec cet outil libre et gratuit. "
-        "Pour une commande publique accessible à toutes et tous."
-    ),
-}
-data_schema = get_data_schema()
-columns = df.columns

src/utils/__init__.py

@@ -0,0 +1,19 @@
+import logging
+import os
+
+logging.basicConfig(
+    format="%(asctime)s %(levelname)-8s %(message)s",
+    level=logging.INFO,
+    datefmt="%Y-%m-%d %H:%M:%S",
+)
+DEVELOPMENT = os.getenv("DEVELOPMENT", "False").lower() == "true"
+logger = logging.getLogger("decp.info")
+
+if DEVELOPMENT:
+    logger.setLevel(logging.DEBUG)
+
+DOMAIN_NAME = (
+    "test.decp.info"
+    if os.getenv("DEVELOPMENT", "False").lower() == "true"
+    else "decp.info"
+)

src/utils/data.py

@@ -0,0 +1,214 @@
+import json
+import logging
+import os
+from collections import OrderedDict
+from datetime import datetime, timedelta
+
+import polars as pl
+from httpx import HTTPError, get
+
+from src.db import get_cursor, schema
+from src.utils import logger
+
+logging.getLogger("httpx").setLevel("WARNING")
+
+
+def get_annuaire_data(siret: str) -> dict:
+    url = f"https://recherche-entreprises.api.gouv.fr/search?q={siret}"
+    try:
+        response = get(url).raise_for_status()
+        response = response.json()["results"][0]
+    except (HTTPError, IndexError):
+        response = None
+        logger.warning("Could not fetch data from recherche-entreprises.api.")
+    return response
+
+
+def get_statistics() -> dict:
+    return (
+        get(
+            "https://www.data.gouv.fr/api/1/datasets/r/0ccf4a75-f3aa-4b46-8b6a-18aeb63e36df",
+            follow_redirects=True,
+        )
+        .raise_for_status()
+        .json()
+    )
+
+
+def get_departements() -> dict:
+    with open("data/departements.json", "rb") as f:
+        data = json.load(f)
+        return data
+
+
+def get_departements_geojson() -> dict:
+    with open("./data/departements-1000m.geojson") as f:
+        geojson = json.load(f)
+
+    # Ajout de feature.id
+    for f in geojson["features"]:
+        f["id"] = f["properties"]["code"]
+
+    return geojson
+
+
+def get_departement_region(code_postal):
+    if code_postal > "97000":
+        code_departement = code_postal[:3]
+    else:
+        code_departement = code_postal[:2]
+    nom_departement = DEPARTEMENTS[code_departement]["departement"]
+    nom_region = DEPARTEMENTS[code_departement]["region"]
+    return code_departement, nom_departement, nom_region
+
+
+def get_data_schema() -> dict:
+    # Récupération du schéma des données tabulaires
+    path = os.getenv("DATA_SCHEMA_PATH")
+    if path.startswith("http"):
+        original_schema: dict = get(
+            os.getenv("DATA_SCHEMA_PATH"), follow_redirects=True
+        ).json()
+    elif os.path.exists(path):
+        with open(path) as f:
+            original_schema: dict = json.load(f)
+    else:
+        raise Exception(f"Chemin vers le schéma invalide: {path}")
+
+    new_schema = OrderedDict()
+
+    for col in original_schema["fields"]:
+        new_schema[col["name"]] = col
+
+    return new_schema
+
+
+def prepare_dashboard_data(
+    lff: pl.LazyFrame,
+    dashboard_year=None,
+    dashboard_acheteur_id=None,
+    dashboard_acheteur_categorie=None,
+    dashboard_acheteur_departement_code=None,
+    dashboard_titulaire_id=None,
+    dashboard_titulaire_categorie=None,
+    dashboard_titulaire_departement_code=None,
+    dashboard_marche_type=None,
+    dashboard_marche_objet=None,
+    dashboard_marche_code_cpv=None,
+    dashboard_marche_considerations_sociales=None,
+    dashboard_marche_considerations_environnementales=None,
+    dashboard_marche_techniques=None,
+    dashboard_marche_innovant=None,
+    dashboard_marche_sous_traitance_declaree=None,
+    dashboard_montant_min=None,
+    dashboard_montant_max=None,
+) -> pl.LazyFrame:
+    if dashboard_year:
+        lff = lff.filter(pl.col("dateNotification").dt.year() == int(dashboard_year))
+    else:
+        lff = lff.filter(
+            pl.col("dateNotification") > (datetime.now() - timedelta(days=365))
+        )
+
+    if dashboard_acheteur_id:
+        lff = lff.filter(pl.col("acheteur_id").str.contains(dashboard_acheteur_id))
+    else:
+        if dashboard_acheteur_categorie:
+            lff = lff.filter(
+                pl.col("acheteur_categorie") == dashboard_acheteur_categorie
+            )
+        if dashboard_acheteur_departement_code:
+            lff = lff.filter(
+                pl.col("acheteur_departement_code").is_in(
+                    dashboard_acheteur_departement_code
+                )
+            )
+
+    if dashboard_titulaire_id:
+        lff = lff.filter(pl.col("titulaire_id").str.contains(dashboard_titulaire_id))
+    else:
+        if dashboard_titulaire_categorie:
+            lff = lff.filter(
+                pl.col("titulaire_categorie") == dashboard_titulaire_categorie
+            )
+        if dashboard_titulaire_departement_code:
+            lff = lff.filter(
+                pl.col("titulaire_departement_code").is_in(
+                    dashboard_titulaire_departement_code
+                )
+            )
+
+    if dashboard_marche_type:
+        lff = lff.filter(pl.col("type") == dashboard_marche_type)
+
+    if dashboard_marche_objet:
+        lff = lff.filter(pl.col("objet").str.contains(f"(?i){dashboard_marche_objet}"))
+
+    if dashboard_marche_code_cpv:
+        lff = lff.filter(pl.col("codeCPV").str.starts_with(dashboard_marche_code_cpv))
+
+    if dashboard_marche_innovant and dashboard_marche_innovant != "all":
+        lff = lff.filter(pl.col("marcheInnovant") == dashboard_marche_innovant)
+
+    if (
+        dashboard_marche_sous_traitance_declaree
+        and dashboard_marche_sous_traitance_declaree != "all"
+    ):
+        lff = lff.filter(
+            pl.col("sousTraitanceDeclaree") == dashboard_marche_sous_traitance_declaree
+        )
+
+    if dashboard_marche_techniques:
+        lff = lff.filter(
+            pl.col("techniques")
+            .str.split(", ")
+            .list.set_intersection(dashboard_marche_techniques)
+            .list.len()
+            > 0
+        )
+
+    if dashboard_marche_considerations_sociales:
+        lff = lff.filter(
+            pl.col("considerationsSociales")
+            .str.split(", ")
+            .list.set_intersection(dashboard_marche_considerations_sociales)
+            .list.len()
+            > 0
+        )
+
+    if dashboard_marche_considerations_environnementales:
+        lff = lff.filter(
+            pl.col("considerationsEnvironnementales")
+            .str.split(", ")
+            .list.set_intersection(dashboard_marche_considerations_environnementales)
+            .list.len()
+            > 0
+        )
+
+    if dashboard_montant_min is not None:
+        lff = lff.filter(pl.col("montant") >= dashboard_montant_min)
+
+    if dashboard_montant_max is not None:
+        lff = lff.filter(pl.col("montant") <= dashboard_montant_max)
+
+    return lff
+
+
+def build_org_frame(org_type: str) -> pl.DataFrame:
+    org_cols = [
+        c
+        for c in schema.names()
+        if c.startswith(f"{org_type}_")
+        and c not in (f"{org_type}_latitude", f"{org_type}_longitude")
+    ]
+    select_list = ", ".join(org_cols)
+    group_list = ", ".join(org_cols)
+    sql = f'SELECT {select_list}, COUNT(*) AS "Marchés" FROM decp GROUP BY {group_list}'
+    return get_cursor().execute(sql).pl()
+
+
+DF_ACHETEURS = build_org_frame("acheteur")
+DF_TITULAIRES = build_org_frame("titulaire")
+DEPARTEMENTS = get_departements()
+DEPARTEMENTS_GEOJSON = get_departements_geojson()
+DATA_SCHEMA = get_data_schema()

src/utils/frontend.py

@@ -0,0 +1,27 @@
+from src.utils.data import DATA_SCHEMA
+
+
+def get_button_properties(height):
+    if height > 65000:
+        download_disabled = True
+        download_text = "Téléchargement désactivé au-delà de 65 000 lignes"
+        download_title = " Ajoutez des filtres pour réduire le nombre de lignes, Excel ne supporte pas d'avoir plus de 65 000 URLs dans une même feuille de calcul."
+    elif height == 0:
+        download_disabled = True
+        download_text = "Pas de données à télécharger"
+        download_title = ""
+    else:
+        download_disabled = False
+        download_text = "Télécharger au format Excel"
+        download_title = "Télécharger les données telles qu'affichées au format Excel"
+    return download_disabled, download_text, download_title
+
+
+def get_enum_values_as_dict(column_name):
+    try:
+        options = {}
+        for value in DATA_SCHEMA[column_name]["enum"]:
+            options[value] = value
+        return options
+    except KeyError:
+        return {"not_found": "not found"}

src/utils/search.py

@@ -0,0 +1,84 @@
+import polars as pl
+from unidecode import unidecode
+
+from src.utils.table import add_links
+from src.utils.tracking import track_search
+
+
+def search_org(dff: pl.DataFrame, query: str, org_type: str) -> pl.DataFrame:
+    """
+    Search in either 'acheteur' or 'titulaire' DataFrame.
+
+    :param dff: Polars DataFrame with acheteur or titulaire columns
+    :param query: User search string
+    :param org_type: 'acheteur' or 'titulaire'
+    :return: Filtered DataFrame with 'matches' column
+    """
+    if not query.strip():
+        return dff.select(pl.lit(False).alias("matches"))
+
+    # Enregistrement des recherche dans Matomo
+    track_search(query, "home_page_search")
+
+    # Normalize query
+    normalized_query = unidecode(query.strip()).upper()
+    tokens = [" " + t.strip() for t in normalized_query.split() if t.strip()]
+
+    # Define columns based on entity type
+    cols = [
+        f"{org_type}_id",
+        f"{org_type}_nom",
+        f"{org_type}_departement_nom",
+        f"{org_type}_departement_code",
+        f"{org_type}_commune_nom",
+    ]
+
+    # Concatenate all fields into one string per row
+    org_str = pl.concat_str(pl.lit(" "), pl.col(cols), separator=" ").str.replace(
+        "-", " "
+    )
+
+    # For each token, create a boolean column: True if token is found
+    token_matches = []
+    for token in tokens:
+        token_match = org_str.str.contains(token).alias(f"token_{token}")
+        token_matches.append(token_match)
+
+    # Count how many tokens match per row
+    match_score = pl.sum_horizontal(token_matches).alias("match_score")
+
+    # For each token, create a boolean column: True if token is found
+    token_matches = []
+    for token in tokens:
+        token_match = org_str.str.contains(token).alias(f"token_{token}")
+        token_matches.append(token_match)
+
+    # Sélection des colonnes
+    if org_type == "acheteur":
+        dff = dff.select(cols + ["Marchés"])
+    if org_type == "titulaire":
+        dff = dff.select(cols + ["Marchés", "titulaire_typeIdentifiant"])
+
+    # Apply and filter
+    dff = (
+        dff.with_columns(token_matches + [match_score])
+        .filter(pl.col("match_score") == len(tokens))
+        .drop([f"token_{token}" for token in tokens])
+    )
+
+    # Format result
+    dff = add_links(dff)
+    dff = dff.with_columns(
+        pl.concat_str(
+            pl.col(f"{org_type}_departement_nom"),
+            pl.lit(" ("),
+            pl.col(f"{org_type}_departement_code"),
+            pl.lit(")"),
+        ).alias("Département")
+    )
+
+    dff = dff.select(f"{org_type}_id", f"{org_type}_nom", "Département", "Marchés")
+    dff = dff.group_by(f"{org_type}_id", f"{org_type}_nom", "Département").sum()
+    dff = dff.sort("Marchés", descending=True)
+
+    return dff

src/utils/seo.py

@@ -0,0 +1,52 @@
+from src.utils import DOMAIN_NAME
+from src.utils.data import get_annuaire_data
+
+
+def make_org_jsonld(org_id, org_type, org_name=None, type_org_id="SIRET") -> dict:
+    org_types = {"acheteur": "GovernmentOrganization", "titulaire": "Organization"}
+    address = None
+    if type_org_id.lower() == "siret" and len(org_id) == 14:
+        annuaire_data = get_annuaire_data(org_id)
+        annuaire_address = annuaire_data["matching_etablissements"][0]
+        code_postal = annuaire_address["code_postal"]
+        commune = annuaire_address["libelle_commune"]
+
+        address = (
+            {
+                "@type": "PostalAddress",
+                "streetAddress": annuaire_address.get("adresse", "")
+                .replace(code_postal, "")
+                .replace(commune, "")
+                .strip(),
+                "addressLocality": commune,
+                "postalCode": code_postal,
+                "addressCountry": "FR",
+            },
+        )
+
+    jsonld = {
+        "@type": org_types[org_type],
+        "name": org_name,
+        "url": f"https://decp.info/{org_type}s/{org_id}",
+        "sameAs": f"https://annuaire-entreprises.data.gouv.fr/etablissement/{org_id}",
+        "identifier": {
+            "@type": "PropertyValue",
+            "propertyID": type_org_id.lower(),
+            "value": org_id,
+        },
+    }
+
+    if address:
+        jsonld["address"] = address
+
+    return jsonld
+
+
+META_CONTENT = {
+    "image_url": f"https://{DOMAIN_NAME}/assets/decp.info.png",
+    "title": "decp.info - exploration des marchés publics français",
+    "description": (
+        "Explorez et analysez les données des marchés publics français avec cet outil libre et gratuit. "
+        "Pour une commande publique accessible à toutes et tous."
+    ),
+}

src/utils/table.py

@@ -0,0 +1,475 @@
+import os
+import uuid
+
+import polars as pl
+from dash import no_update
+from polars import selectors as cs
+
+from src.db import query_marches, schema
+from src.utils import logger
+from src.utils.data import DATA_SCHEMA
+from src.utils.frontend import get_button_properties
+from src.utils.tracking import track_search
+
+
+def split_filter_part(filter_part):
+    operators = [
+        ["s<", "<"],
+        ["s>", ">"],
+        ["i<", "<"],
+        ["i>", ">"],
+        ["icontains", "contains"],
+        # [" ", "contains"]
+    ]
+    logger.debug("filter part " + filter_part)
+    for operator_group in operators:
+        if operator_group[0] in filter_part:
+            name_part, value_part = filter_part.split(operator_group[0], 1)
+            name_part = name_part.strip()
+            value = value_part.strip()
+            name = name_part[name_part.find("{") + 1 : name_part.rfind("}")]
+            logger.debug("=> " + " ".join([name, operator_group[1], value]))
+
+            return name, operator_group[1], value
+
+    return [None] * 3
+
+
+def add_resource_link(dff: pl.DataFrame) -> pl.DataFrame:
+    dff = dff.with_columns(
+        (
+            '<a href="' + pl.col("sourceFile") + '">' + pl.col("sourceDataset") + "</a>"
+        ).alias("sourceDataset")
+    )
+    dff = dff.drop(["sourceFile"])
+    return dff
+
+
+def add_links(dff: pl.DataFrame):
+    for col in ["uid", "acheteur_nom", "titulaire_nom", "acheteur_id", "titulaire_id"]:
+        if col in dff.columns:
+            if col.startswith("titulaire_"):
+                detail_link = (
+                    '<a href = "/titulaires/'
+                    + pl.col("titulaire_id")
+                    + '">'
+                    + pl.col(col)
+                    + "</a>"
+                )
+                if col == "titulaire_nom":
+                    detail_link = (
+                        detail_link
+                        + ' <a href="/observatoire?titulaire_id='
+                        + pl.col("titulaire_id")
+                        + '" title="Voir dans l\'observatoire">📊</a>'
+                    )
+                dff = dff.with_columns(
+                    pl.when(
+                        pl.Expr.or_(
+                            pl.col("titulaire_typeIdentifiant").is_null(),
+                            pl.col("titulaire_typeIdentifiant") == "SIRET",
+                        )
+                    )
+                    .then(detail_link)
+                    .otherwise(pl.col(col))
+                    .alias(col)
+                )
+            if col.startswith("acheteur_"):
+                detail_link = (
+                    '<a href = "/acheteurs/'
+                    + pl.col("acheteur_id")
+                    + '">'
+                    + pl.col(col)
+                    + "</a>"
+                )
+                if col == "acheteur_nom":
+                    detail_link = (
+                        detail_link
+                        + ' <a href="/observatoire?acheteur_id='
+                        + pl.col("acheteur_id")
+                        + '" title="Voir dans l\'observatoire">📊</a>'
+                    )
+                dff = dff.with_columns(detail_link.alias(col))
+            if col == "uid":
+                dff = dff.with_columns(
+                    (
+                        '<a href = "/marches/'
+                        + pl.col("uid")
+                        + '">'
+                        + pl.col("uid")
+                        + "</a>"
+                    ).alias("uid")
+                )
+    return dff
+
+
+def add_links_in_dict(data: list[dict], org_type: str) -> list:
+    new_data = []
+    for marche in data:
+        org_id = marche[org_type + "_id"]
+        marche[org_type + "_nom"] = (
+            f'<a href="/{org_type}s/{org_id}">{marche[org_type + "_nom"]}</a>'
+        )
+        if marche.get("uid"):
+            marche["id"] = f'<a href="/marches/{marche["uid"]}">{marche["id"]}</a>'
+            marche["uid"] = f'<a href="/marches/{marche["uid"]}">{marche["uid"]}</a>'
+        new_data.append(marche)
+    return new_data
+
+
+def booleans_to_strings(lff: pl.LazyFrame) -> pl.LazyFrame:
+    """
+    Convert all boolean columns to string type.
+    """
+    lff = lff.with_columns(
+        pl.col(cs.Boolean)
+        .cast(pl.String)
+        .str.replace("true", "oui")
+        .str.replace("false", "non")
+    )
+    return lff
+
+
+def numbers_to_strings(lff: pl.LazyFrame) -> pl.LazyFrame:
+    """
+    Convert all numeric columns to string type.
+    """
+    lff = lff.with_columns(pl.col(pl.Float64, pl.Int16).cast(pl.String).fill_null(""))
+    return lff
+
+
+def dates_to_strings(lff: pl.LazyFrame, column: str) -> pl.LazyFrame:
+    """
+    Convert a date column to string type.
+    """
+    lff = lff.with_columns(pl.col(column).cast(pl.String).fill_null(""))
+    return lff
+
+
+def format_number(number) -> str:
+    number = "{:,}".format(number).replace(",", " ")
+    return number
+
+
+def unformat_montant(number: str) -> float:
+    number = number.replace(" €", "")
+    number = number.replace(" €", "").replace(" ", "")
+    number = number.replace(",", ".")
+    number = number.strip()
+    return float(number)
+
+
+def format_values(dff: pl.DataFrame) -> pl.DataFrame:
+    def format_montant(expr, scale=None):
+        # https://stackoverflow.com/a/78636786
+        expr = expr.cast(pl.String)
+        expr = expr.str.splitn(".", 2)
+
+        num = expr.struct[0]
+        frac = expr.struct[1]
+
+        # Ajout des espaces
+        num = (
+            num.str.reverse()
+            .str.replace_all(r"\d{3}", "$0 ")
+            .str.reverse()
+            .str.replace(r"^ ", "")
+        )
+
+        frac: pl.Expr = (
+            pl.when(frac.is_not_null() & ~frac.is_in(["0"]))
+            .then("," + frac.str.head(2))
+            .otherwise(pl.lit(""))
+        )
+
+        montant: pl.Expr = (
+            pl.when((num + frac) == pl.lit(""))
+            .then(pl.lit(""))
+            .otherwise(num + frac + pl.lit(" €"))
+        )
+
+        return montant
+
+    def format_distance(expr):
+        expr = expr.cast(pl.String)
+        return pl.concat_str(expr, pl.lit(" km"))
+
+    if "montant" in dff.columns:
+        dff = dff.with_columns(pl.col("montant").pipe(format_montant).alias("montant"))
+    if "titulaire_distance" in dff.columns:
+        dff = dff.with_columns(
+            pl.col("titulaire_distance")
+            .pipe(format_distance)
+            .alias("titulaire_distance")
+        )
+
+    return dff
+
+
+def filter_table_data(
+    lff: pl.LazyFrame, filter_query: str, filter_source: str
+) -> pl.LazyFrame:
+    _schema = lff.collect_schema()
+    track_search(filter_query, filter_source)
+    filtering_expressions = filter_query.split(" && ")
+    for filter_part in filtering_expressions:
+        col_name, operator, filter_value = split_filter_part(filter_part)
+        col_type = str(_schema[col_name])
+        # logger.debug("filter_value:", filter_value)
+        # logger.debug("filter_value_type:", type(filter_value))
+        # logger.debug("operator:", operator)
+        # logger.debug("col_type:", col_type)
+
+        lff = lff.filter(pl.col(col_name).is_not_null())
+
+        if col_type == "Date":
+            # Convertir la colonne date en chaînes de caractères
+            lff = dates_to_strings(lff, col_name)
+            col_type = "String"
+        if col_type == "String":
+            lff = lff.filter(pl.col(col_name) != pl.lit(""))
+
+        elif col_type.startswith("Int") or col_type.startswith("Float"):
+            try:
+                filter_value = int(filter_value)
+            except ValueError:
+                logger.error(f"Invalid numeric filter value: {filter_value}")
+                continue
+
+        if operator in ("contains", "<", "<=", ">", ">="):
+            if operator == "<":
+                lff = lff.filter(pl.col(col_name) < filter_value)
+            elif operator == ">":
+                lff = lff.filter(pl.col(col_name) > filter_value)
+            elif operator == ">=":
+                lff = lff.filter(pl.col(col_name) >= filter_value)
+            elif operator == "<=":
+                lff = lff.filter(pl.col(col_name) <= filter_value)
+            elif operator == "contains":
+                if col_type in ["String", "Date"]:
+                    filter_value = filter_value.strip('"')
+                    if filter_value.endswith("*"):
+                        lff = lff.filter(
+                            pl.col(col_name).str.starts_with(filter_value[:-1])
+                        )
+                    elif filter_value.startswith("*"):
+                        lff = lff.filter(
+                            pl.col(col_name).str.ends_with(filter_value[1:])
+                        )
+                    else:
+                        lff = lff.filter(
+                            pl.col(col_name).str.contains("(?i)" + filter_value)
+                        )
+                elif col_type.startswith("Int") or col_type.startswith("Float"):
+                    lff = lff.filter(pl.col(col_name) == filter_value)
+                else:
+                    logger.error(f"Invalid column type: {col_type}")
+        else:
+            logger.error(f"Invalid operator: {operator}")
+
+        # elif operator == 'datestartswith':
+        # lff = lff.filter(pl.col(col_name).str.startswith(filter_value)")
+
+    return lff
+
+
+def sort_table_data(lff: pl.LazyFrame, sort_by: list) -> pl.LazyFrame:
+    lff = lff.sort(
+        [col["column_id"] for col in sort_by],
+        descending=[col["direction"] == "desc" for col in sort_by],
+        nulls_last=True,
+    )
+    logger.debug(sort_by)
+    return lff
+
+
+def setup_table_columns(
+    dff, hideable: bool = True, exclude: list = None, new_columns: list = None
+) -> tuple:
+    # Liste finale de colonnes
+    markdown_exceptions = ["montant", "titulaire_distance", "distance", "dureeMois"]
+    columns = []
+    tooltip = {}
+    for column_id in dff.columns:
+        if exclude and column_id in exclude:
+            continue
+        column_object = DATA_SCHEMA.get(column_id)
+        if column_object:
+            column_name = column_object.get("title")
+        else:
+            # Si le champ est un champ créé par erreur lors d'une jointure, on le skip
+            if column_id.endswith("_left") or column_id.endswith("_right"):
+                logger.warning(f"Champ innatendu : {column_id}")
+                continue
+            column_name = column_id
+            column_object = {"title": column_name, "description": ""}
+
+        presentation = "input" if column_id in markdown_exceptions else "markdown"
+
+        column = {
+            "name": column_name,
+            "id": column_id,
+            "presentation": presentation,
+            "type": "text",
+            "format": {"nully": "N/A"},
+            "hideable": hideable,
+        }
+        columns.append(column)
+        if column_object:
+            tooltip[column_id] = {
+                "value": f"""**{column_object.get("title")}** ({column_id})
+
+    """
+                + column_object.get("description", ""),
+                "type": "markdown",
+            }
+    return columns, tooltip
+
+
+def get_default_hidden_columns(page):
+    if page == "acheteur":
+        displayed_columns = [
+            "uid",
+            "objet",
+            "dateNotification",
+            "titulaire_id",
+            "titulaire_typeIdentifiant",
+            "titulaire_nom",
+            "titulaire_distance",
+            "montant",
+            "codeCPV",
+            "dureeRestanteMois",
+        ]
+    elif page == "titulaire":
+        displayed_columns = [
+            "uid",
+            "objet",
+            "dateNotification",
+            "acheteur_id",
+            "acheteur_nom",
+            "titulaire_distance",
+            "montant",
+            "codeCPV",
+            "dureeRestanteMois",
+        ]
+    elif page == "tableau":
+        displayed_columns = os.getenv("DISPLAYED_COLUMNS")
+    else:
+        displayed_columns = os.getenv("DISPLAYED_COLUMNS")
+        logger.warning(f"Invalid page: {page}")
+
+    hidden_columns = []
+
+    for col in schema.names():
+        if col in displayed_columns:
+            continue
+        else:
+            hidden_columns.append(col)
+    return hidden_columns
+
+
+def prepare_table_data(
+    data, data_timestamp, filter_query, page_current, page_size, sort_by, source_table
+):
+    """
+    Fonction de préparation des données pour les datatables, afin de permettre une gestion fine des logiques,
+    notamment pour les filtres et les tris.
+    :param data
+    :param data_timestamp:
+    :param filter_query:
+    :param page_current:
+    :param page_size:
+    :param sort_by:
+    :param source_table:
+    :return:
+    """
+
+    if os.getenv("DEVELOPMENT").lower() == "true":
+        logger.debug(" + + + + + + + + + + + + + + + + + + ")
+
+    trigger_cleanup = no_update
+
+    # Récupération des données
+    if isinstance(data, list):
+        lff: pl.LazyFrame = pl.LazyFrame(data, strict=False, infer_schema_length=5000)
+    elif isinstance(data, pl.LazyFrame):
+        lff = data
+    else:
+        lff: pl.LazyFrame = query_marches().lazy()
+
+    # Application des filtres
+    if filter_query:
+        lff = filter_table_data(lff, filter_query, source_table)
+        trigger_cleanup = no_update if source_table == "tableau" else str(uuid.uuid4())
+
+    # Application des tris
+    if sort_by and len(sort_by) > 0:
+        lff = sort_table_data(lff, sort_by)
+
+    # Matérialisation des filtres
+    dff: pl.DataFrame = lff.collect()
+    height = dff.height
+
+    if height > 0:
+        nb_rows = f"{format_number(height)} lignes ({format_number(dff.select('uid').unique().height)} marchés)"
+    else:
+        nb_rows = "0 lignes (0 marchés)"
+
+    # Pagination des données
+    start_row = page_current * page_size
+    # end_row = (page_current + 1) * page_size
+    dff = dff.slice(start_row, page_size)
+
+    # Tout devient string
+    dff = dff.cast(pl.String)
+
+    # Remplace les strings null par "", mais pas les numeric null
+    dff = dff.fill_null("")
+
+    # Ajout des liens vers les pages de détails
+    dff = add_links(dff)
+
+    # Ajout des liens vers les fichiers Open Data
+    if "sourceFile" in dff.columns:
+        dff = add_resource_link(dff)
+
+    # Formatage des montants
+    if height > 0:
+        dff = format_values(dff)
+
+    # Récupération des colonnes et tooltip
+    table_columns, tooltip = setup_table_columns(dff)
+
+    dicts = dff.to_dicts()
+
+    # Propriétés du bouton de téléchargement
+    download_disabled, download_text, download_title = get_button_properties(height)
+
+    return (
+        dicts,
+        table_columns,
+        tooltip,
+        data_timestamp + 1,
+        nb_rows,
+        download_disabled,
+        download_text,
+        download_title,
+        trigger_cleanup,
+    )
+
+
+def invert_columns(columns):
+    """
+    Renvoie les colonnes du schéma non spécifiées en paramètre. Utile pour passer d'une colonnes masquées à une liste de colonnes affichées, et vice versa.
+
+    :param columns:
+    :return:
+    """
+    inverted_columns = []
+    for column in schema.names():
+        if column not in columns:
+            inverted_columns.append(column)
+    return inverted_columns
+
+
+COLUMNS = schema.names()

src/utils/tracking.py

@@ -0,0 +1,30 @@
+import os
+import uuid
+from time import localtime
+
+from httpx import post
+
+from src.utils import DEVELOPMENT
+
+
+def track_search(query, category):
+    if len(query) >= 4 and not DEVELOPMENT and os.getenv("MATOMO_DOMAIN"):
+        url = "https://decp.info"
+        params = {
+            "idsite": os.getenv("MATOMO_ID_SITE"),
+            "url": url,
+            "rec": "1",
+            "action_name": "search" if category == "home_page_search" else "filter",
+            "search_cat": category,
+            "rand": uuid.uuid4().hex,
+            "apiv": "1",
+            "h": localtime().tm_hour,
+            "m": localtime().tm_min,
+            "s": localtime().tm_sec,
+            "search": query,
+            "token_auth": os.getenv("MATOMO_TOKEN"),
+        }
+        post(
+            url=f"https://{os.getenv('MATOMO_DOMAIN')}/matomo.php",
+            params=params,
+        ).raise_for_status()

tests/conftest.py

@@ -1,5 +1,6 @@
 import datetime
 import os
+from pathlib import Path
 
 import polars as pl
 import pytest
@@ -41,12 +42,19 @@ def test_data():
             "titulaire_categorie": "PME",
         }
     ]
-    path = "tests/test.parquet"
+    parquet_path = Path(os.path.abspath("tests/test.parquet"))
-    path = os.path.abspath(path)
+    db_path = parquet_path.parent / "decp.duckdb"
-    print(f"Writing test data to: {path}")  # <-- This will show you the real path
+    print(f"Writing test data to: {parquet_path}")
 
-    pl.DataFrame(data).write_parquet("tests/test.parquet")
+    pl.DataFrame(data).write_parquet(parquet_path)
-    yield path
+
+    # Remove any stale DuckDB from a previous run so src.db rebuilds from
+    # the freshly-written parquet at import time.
+    for artifact in (db_path, db_path.with_suffix(".duckdb.tmp")):
+        if artifact.exists():
+            artifact.unlink()
+
+    yield str(parquet_path)
 
 
 def pytest_setup_options():

tests/test_db.py

@@ -0,0 +1,267 @@
+import datetime
+import os
+import time
+
+import polars as pl
+import pytest
+
+from src.db import should_rebuild
+
+
+@pytest.fixture
+def parquet_and_db(tmp_path, monkeypatch):
+    parquet = tmp_path / "source.parquet"
+    db = tmp_path / "decp.duckdb"
+    parquet.write_bytes(b"fake parquet content")
+    monkeypatch.delenv("REBUILD_DUCKDB", raising=False)
+    monkeypatch.delenv("DEVELOPMENT", raising=False)
+    return parquet, db
+
+
+def test_should_rebuild_when_db_missing(parquet_and_db):
+    parquet, db = parquet_and_db
+    assert should_rebuild(db, parquet) is True
+
+
+def test_should_rebuild_prod_when_parquet_newer(parquet_and_db, monkeypatch):
+    parquet, db = parquet_and_db
+    db.write_bytes(b"x")
+    parquet.touch()
+    now = time.time()
+    os.utime(db, (now, now))
+    os.utime(parquet, (now + 10, now + 10))
+    monkeypatch.setenv("DEVELOPMENT", "false")
+    assert should_rebuild(db, parquet) is True
+
+
+def test_should_not_rebuild_prod_when_parquet_older(parquet_and_db, monkeypatch):
+    parquet, db = parquet_and_db
+    parquet.touch()
+    db.write_bytes(b"x")
+    now = time.time()
+    os.utime(parquet, (now, now))
+    os.utime(db, (now + 10, now + 10))
+    monkeypatch.setenv("DEVELOPMENT", "false")
+    assert should_rebuild(db, parquet) is False
+
+
+def test_should_not_rebuild_dev_even_when_parquet_newer(parquet_and_db, monkeypatch):
+    parquet, db = parquet_and_db
+    db.write_bytes(b"x")
+    parquet.touch()
+    now = time.time()
+    os.utime(db, (now, now))
+    os.utime(parquet, (now + 10, now + 10))
+    monkeypatch.setenv("DEVELOPMENT", "true")
+    monkeypatch.delenv("REBUILD_DUCKDB", raising=False)
+    assert should_rebuild(db, parquet) is False
+
+
+def test_should_rebuild_dev_when_rebuild_forced(parquet_and_db, monkeypatch):
+    parquet, db = parquet_and_db
+    db.write_bytes(b"x")
+    parquet.touch()
+    now = time.time()
+    os.utime(db, (now, now))
+    os.utime(parquet, (now + 10, now + 10))
+    monkeypatch.setenv("DEVELOPMENT", "true")
+    monkeypatch.setenv("REBUILD_DUCKDB", "true")
+    assert should_rebuild(db, parquet) is True
+
+
+@pytest.fixture
+def built_db(tmp_path, monkeypatch):
+    """Build a DuckDB from a small Polars frame written as parquet."""
+    parquet_path = tmp_path / "source.parquet"
+    db_path = tmp_path / "decp.duckdb"
+
+    data = pl.DataFrame(
+        [
+            {
+                "uid": "1",
+                "id": "1",
+                "objet": "Travaux",
+                "acheteur_id": "123",
+                "acheteur_nom": "ACHETEUR 1",
+                "acheteur_departement_code": "75",
+                "acheteur_departement_nom": "Paris",
+                "acheteur_commune_nom": "Paris",
+                "titulaire_commune_nom": "Paris",
+                "titulaire_departement_nom": "Paris",
+                "titulaire_id": "345",
+                "titulaire_nom": "TITULAIRE 1",
+                "titulaire_departement_code": "35",
+                "titulaire_typeIdentifiant": "SIRET",
+                "montant": 1000.0,
+                "dateNotification": datetime.date(2025, 1, 1),
+                "donneesActuelles": True,
+                "marcheInnovant": True,
+            },
+            {
+                "uid": "2",
+                "id": "2",
+                "objet": "Études",
+                "acheteur_id": "123",
+                "acheteur_nom": "ACHETEUR 1",
+                "acheteur_departement_code": "75",
+                "acheteur_departement_nom": "Paris",
+                "acheteur_commune_nom": "Paris",
+                "titulaire_commune_nom": "Paris",
+                "titulaire_departement_nom": "Paris",
+                "titulaire_id": "567",
+                "titulaire_nom": None,
+                "titulaire_departement_code": "75",
+                "titulaire_typeIdentifiant": "SIRET",
+                "montant": 500.0,
+                "dateNotification": datetime.date(2024, 6, 1),
+                "donneesActuelles": True,
+                "marcheInnovant": False,
+            },
+            {
+                "uid": "3",
+                "id": "3",
+                "objet": "Ancien",
+                "acheteur_id": "A2",
+                "acheteur_nom": None,
+                "acheteur_departement_code": "13",
+                "acheteur_departement_nom": "Paris",
+                "acheteur_commune_nom": "Paris",
+                "titulaire_commune_nom": "Paris",
+                "titulaire_departement_nom": "Paris",
+                "titulaire_id": "T3",
+                "titulaire_nom": "Autre",
+                "titulaire_departement_code": "13",
+                "titulaire_typeIdentifiant": "SIRET",
+                "montant": 100.0,
+                "dateNotification": datetime.date(2023, 1, 1),
+                "donneesActuelles": False,  # must be filtered out
+                "marcheInnovant": False,
+            },
+        ]
+    )
+    data.write_parquet(parquet_path)
+    monkeypatch.setenv("DATA_FILE_PARQUET_PATH", str(parquet_path))
+
+    from src.db import build_database
+
+    build_database(db_path, parquet_path)
+    return db_path
+
+
+def test_build_filters_donnees_actuelles(built_db):
+    import duckdb
+
+    with duckdb.connect(str(built_db), read_only=True) as c:
+        rows = c.execute("SELECT uid FROM decp ORDER BY uid").fetchall()
+    assert [r[0] for r in rows] == ["1", "2"]
+
+
+def test_build_converts_booleans_to_oui_non(built_db):
+    import duckdb
+
+    with duckdb.connect(str(built_db), read_only=True) as c:
+        values = c.execute("SELECT marcheInnovant FROM decp ORDER BY uid").fetchall()
+    assert [v[0] for v in values] == ["oui", "non"]
+
+
+def test_build_replaces_null_org_names(built_db):
+    import duckdb
+
+    with duckdb.connect(str(built_db), read_only=True) as c:
+        titulaire_2 = c.execute(
+            "SELECT titulaire_nom FROM decp WHERE uid = '2'"
+        ).fetchone()
+    assert titulaire_2[0] == "[Identifiant non reconnu dans la base INSEE]"
+
+
+def test_build_creates_derived_tables(built_db):
+    import duckdb
+
+    with duckdb.connect(str(built_db), read_only=True) as c:
+        tables = {r[0] for r in c.execute("SHOW TABLES").fetchall()}
+    assert {
+        "decp",
+        "acheteurs_marches",
+        "titulaires_marches",
+        "acheteurs_departement",
+        "titulaires_departement",
+    } <= tables
+
+
+def test_query_marches_returns_polars_frame(built_db, monkeypatch):
+    monkeypatch.setenv(
+        "DATA_FILE_PARQUET_PATH", str(built_db.parent / "source.parquet")
+    )
+    # Force src.db to load pointing at this test DB.
+    import importlib
+
+    import src.db
+
+    importlib.reload(src.db)
+    from src.db import query_marches
+
+    frame = query_marches("acheteur_id = ?", ("123",))
+    assert isinstance(frame, pl.DataFrame)
+    assert frame.height == 2
+    assert set(frame["uid"].to_list()) == {"1", "2"}
+
+
+def test_concurrent_build_serialized(tmp_path):
+    """Multiple threads calling _ensure_database must serialize via flock.
+
+    Only one should actually build; others wait, see the fresh DB, and skip.
+    No tmp file should leak. No exceptions should occur.
+    """
+    import fcntl
+    import threading
+
+    import src.db as db
+
+    # Set up source parquet
+    parquet_path = tmp_path / "src.parquet"
+    df = pl.DataFrame(
+        {
+            "uid": ["A"],
+            "donneesActuelles": [True],
+            "dateNotification": ["2024-01-01"],
+            "objet": ["Test"],
+            "acheteur_id": ["a1"],
+            "acheteur_nom": ["A1"],
+            "titulaire_id": ["t1"],
+            "titulaire_nom": ["T1"],
+            "acheteur_departement_code": ["75"],
+            "titulaire_departement_code": ["75"],
+            "montant": [1000.0],
+            "dureeMois": [12],
+        }
+    )
+    df.write_parquet(parquet_path)
+
+    db_path = tmp_path / "decp.duckdb"
+    lock_path = db_path.with_suffix(".duckdb.lock")
+    tmp_path_artifact = db_path.with_suffix(".duckdb.tmp")
+
+    errors: list[BaseException] = []
+
+    def worker():
+        try:
+            # Mirror the locking logic in _ensure_database
+            with open(lock_path, "w") as lf:
+                fcntl.flock(lf.fileno(), fcntl.LOCK_EX)
+                try:
+                    if db.should_rebuild(db_path, parquet_path):
+                        db.build_database(db_path, parquet_path)
+                finally:
+                    fcntl.flock(lf.fileno(), fcntl.LOCK_UN)
+        except BaseException as exc:
+            errors.append(exc)
+
+    threads = [threading.Thread(target=worker) for _ in range(3)]
+    for t in threads:
+        t.start()
+    for t in threads:
+        t.join()
+
+    assert errors == []
+    assert db_path.exists()
+    assert not tmp_path_artifact.exists()

tests/test_main.py

@@ -52,7 +52,6 @@ def test_002_filter_persistence(dash_duo: DashComposite):
         return _filter_input
 
     for page in ["tableau", "acheteurs/123", "titulaires/345"]:
-        print("page:", page)
         filter_input = open_page_and_check_filter_input()
         filter_input.send_keys("11")  # a UID that doesn't exist
         filter_input.send_keys(Keys.ENTER)
@@ -91,7 +90,7 @@ def test_003_tableau_download(dash_duo: DashComposite):
 def test_004_add_links_observatoire_acheteur():
     import polars as pl
 
-    from src.utils import add_links
+    from src.utils.table import add_links
 
     dff = pl.DataFrame(
         {
@@ -116,7 +115,7 @@ def test_004_add_links_observatoire_acheteur():
 def test_005_add_links_observatoire_titulaire():
     import polars as pl
 
-    from src.utils import add_links
+    from src.utils.table import add_links
 
     dff = pl.DataFrame(
         {
@@ -219,9 +218,7 @@ def test_008_search_to_observatoire(dash_duo: DashComposite):
 def test_010_observatoire_montant_filter():
     import datetime
 
-    import polars as pl
+    from src.utils.data import prepare_dashboard_data
-
-    from src.utils import prepare_dashboard_data
 
     data = pl.DataFrame(
         {