Filtering data further

README.md

@@ -26,7 +26,7 @@ GoodDocs-v0 is a text dataset scraped from high-quality documentation sources in
 - `awesome-repos.parquet` — structured links extracted from Awesome lists-of-lists (`name`, `link`, `description`, `source_repo`, optional `stars`).
 - `data_collection_utils/` — utilities to regenerate the dataset:
   - `scrape_gh_docs.py` — main scraper/collector for documentation from GitHub repositories.
-  - `parse_gh_docs_config.yaml` — reproducible configuration (inputs, outputs, filters, strategies).
+  - `scrape_gh_docs_config.yaml` — reproducible configuration (inputs, outputs, filters, strategies).
   - `github_links.txt` — the seed list of GitHub repositories (e.g., top repositories by stars).
   - `awesome_final_repos.py` — extractor for non-"awesome" repositories referenced by Awesome lists.
   - `awesome_scrap_config.yaml` — configuration for `awesome_final_repos.py` (root, depth, output, cache, workers, optional `fetch_stars`).
@@ -63,7 +63,7 @@ Typical uses:
 
 ## Reproducing the dataset
 
-The scraper is configurable and designed to be reproducible via `data_collection_utils/parse_gh_docs_config.yaml`.
+The scraper is configurable and designed to be reproducible via `data_collection_utils/scrape_gh_docs_config.yaml`.
 
 1) Prerequisites
    - System tools: `git`
@@ -83,7 +83,7 @@ python3 data_collection_utils/scrape_gh_docs.py
 python3 data_collection_utils/scrape_gh_docs.py --no-fetch
 ```
 
-Configuration (YAML-driven; see `data_collection_utils/parse_gh_docs_config.yaml`):
+Configuration (YAML-driven; see `data_collection_utils/scrape_gh_docs_config.yaml`):
 
 - `input` — path to a file containing one repo per line (owner/repo or full URL)
 - `outdir`, `md_failed`, `texts_parquet`

clean/clean_meta.yaml

@@ -12,7 +12,7 @@ out_filtered_reasons_csv: ../output/repometa.filtered_out.csv
 # Keep only repos whose primaryLanguage is in this list (empty means no include filter)
 include_languages: []
 # Exclude repos whose primaryLanguage is in this list
-exclude_languages: [null] # filter empty here.
+exclude_languages: [null] # filter empty values here.
 # Minimum number of stars
 min_stars: 300
 # Exclude forks
@@ -25,4 +25,4 @@ include_owners: []
 exclude_owners: []
 # Topic filters (substring match, case-insensitive) over comma-joined topics field
 include_topic_substrings: []
-exclude_topic_substrings: ["interview","interview-prep","learn"]
+exclude_topic_substrings: ["interview","interview-prep","learn","roadmap","chinese"]

data_collection_utils/fetch_gh_meta.py

@@ -27,6 +27,7 @@ import pandas as pd
 import yaml
 from tqdm import tqdm
 import logging
+from datetime import datetime
 
 from github_api_utils import fetch_repos_metadata_graphql
 
@@ -107,9 +108,8 @@ def main():
         return [_resolve_cfg_path(val)]
 
     input_parquet_values = _resolve_cfg_paths(cfg.get("input_parquet"))
-    out_parquet_value = _resolve_cfg_path(
-        cfg.get("out_parquet", "../output/repometa.parquet")
-    )
+    out_parquet_value = _resolve_cfg_path(cfg.get("out_parquet", "../output/repometa.parquet"))
+    resume = bool(cfg.get("resume", True))
     batch_size = int(cfg.get("batch_size", 20))
     quiet = bool(cfg.get("quiet", False))
 
@@ -139,10 +139,23 @@ def main():
             seen.add(key)
             pairs.append((owner, repo))
 
-    logger.info(f"Total unique repos to fetch: {len(pairs)}")
+    # Resume: if output exists and resume=true, skip already-present repos
+    existing_map = {}
+    out_path = Path(out_parquet_value)
+    if resume and out_path.exists():
+        try:
+            existing_df = pd.read_parquet(out_path)
+            if {"owner", "repo"}.issubset(existing_df.columns):
+                existing_map = {f"{o}/{r}": True for o, r in zip(existing_df["owner"], existing_df["repo"]) }
+        except Exception:
+            existing_map = {}
+    if existing_map:
+        pairs = [(o, r) for (o, r) in pairs if f"{o}/{r}" not in existing_map]
+    logger.info(f"Total unique repos to fetch: {len(pairs)} (resume={'on' if resume else 'off'})")
 
     # Fetch in batches via GraphQL
     records: List[Dict[str, Any]] = []
+    run_ts = datetime.utcnow().isoformat()
     for i in tqdm(range(0, len(pairs), batch_size), desc="GraphQL batches"):
         batch = pairs[i : i + batch_size]
         meta = fetch_repos_metadata_graphql(batch)
@@ -168,13 +181,29 @@ def main():
                     ),
                     "is_fork": m.get("is_fork"),
                     "parent_url": m.get("parent_url"),
+                    "updated_at": run_ts,
                 }
             )
 
     df_out = pd.DataFrame(records)
     out_path = Path(out_parquet_value)
     out_path.parent.mkdir(parents=True, exist_ok=True)
-    df_out.to_parquet(out_path)
+    if resume and out_path.exists():
+        try:
+            existing_df = pd.read_parquet(out_path)
+            # Ensure updated_at exists on existing_df as well
+            if "updated_at" not in existing_df.columns:
+                existing_df["updated_at"] = None
+            combined = pd.concat([existing_df, df_out], ignore_index=True)
+            # Drop duplicates by owner/repo keeping last (newest fetch)
+            combined = combined.drop_duplicates(subset=["owner", "repo"], keep="last")
+            combined.to_parquet(out_path, index=False)
+            logger.info(f"Appended {len(df_out)} new repos (resume) to {out_path} (total {len(combined)})")
+            return
+        except Exception:
+            # If any issue, fall back to overwrite with new
+            pass
+    df_out.to_parquet(out_path, index=False)
     logger.info(f"Wrote metadata for {len(df_out)} repos to {out_path}")
 
 

data_collection_utils/fetch_gh_meta_config.yaml

@@ -12,3 +12,6 @@ batch_size: 20
 
 # Logging
 quiet: false
+
+# Resume: skip refetching repos that already exist in out_parquet
+resume: true

data_collection_utils/scrape_gh_docs.py

@@ -14,14 +14,14 @@ Key features:
     3) Zip fallback (optional): `--prefer-zip` to download a codeload zip (no REST usage) and extract only .md.
     4) Org heuristics and search fallback via GitHub API if direct docs folder not found.
  - Content selection: `--only-md` limits downloads/extractions to Markdown files.
- - Central config: reads YAML from `parse_gh_docs_config.yaml` to control inputs/outputs and strategies.
+ - Central config: reads YAML from `scrape_gh_docs_config.yaml` to control inputs/outputs and strategies.
  - Note: Repository metadata fetching and filtering (e.g., by age/language/topics) has been split
    into a separate pipeline step (see `data_collection_utils/fetch_gh_meta.py` and `clean/clean_meta.py`).
  - Quiet mode: `--quiet` or YAML `quiet: true` switches logging to warnings+ so tqdm progress stays visible.
  - No-fetch mode: `--no-fetch` rebuilds Parquet(s) from existing outdir without any network calls. You can also emit a per-file texts Parquet via `--texts-parquet` or YAML `texts_parquet`.
 
 Typical usage:
-    uv run starting_data/scrape_gh_docs.py --config starting_data/parse_gh_docs_config.yaml
+    uv run starting_data/scrape_gh_docs.py --config starting_data/scrape_gh_docs_config.yaml
 
 Outputs:
  - Saves files under `<outdir>/<owner>__<repo>/...`.
@@ -48,6 +48,7 @@ import subprocess
 import yaml
 import duckdb
 import logging
+from datetime import datetime
 import langid  # https://github.com/saffsd/langid.py
 
 
@@ -145,11 +146,12 @@ def collect_md_rows_for_repo_dir(
     outdir: Path,
     lang_filter_value: Optional[str],
     min_text_chars_value: int,
+    updated_at: str,
 ) -> List[Dict[str, Any]]:
     """Scan a single <owner>__<repo> directory for Markdown files and build row dicts.
 
     Returns a list of rows with fields: owner, repo, repo_dir, file_rel_repo,
-    file_rel_outdir, size, mtime, lang, content.
+    file_rel_outdir, size, mtime, lang, content, updated_at.
     """
     try:
         owner, repo = d.name.split("__", 1)
@@ -185,6 +187,7 @@ def collect_md_rows_for_repo_dir(
                 "mtime": int(md_file.stat().st_mtime),
                 "lang": lang_code,
                 "content": text,
+                "updated_at": updated_at,
             }
             rows.append(row)
     return rows
@@ -529,6 +532,7 @@ def process_repo_entry(
     prefer_zip: bool = False,
     prefer_sparse: bool = False,
     only_md: bool = False,
+    resume: bool = True,
 ):
     owner_repo = owner_repo.strip()
     if not owner_repo or owner_repo.startswith("#"):
@@ -557,6 +561,34 @@ def process_repo_entry(
     got_any = False
     default_branch = None
 
+    # Resume: if repo directory already exists, skip network fetch and use existing files
+    repo_saved_root = outdir / safe_name(f"{owner}__{repo}")
+    if resume and repo_saved_root.exists():
+        # Determine docs folder similar to below logic
+        if (repo_saved_root / "docs").exists():
+            docs_folder = repo_saved_root / "docs"
+        else:
+            found = None
+            for p in repo_saved_root.rglob("docs"):
+                if p.is_dir():
+                    found = p
+                    break
+            docs_folder = found if found else repo_saved_root
+        md_count = count_md_files(docs_folder)
+        result["default_branch"] = None
+        result["method"] = "resume-existing"
+        result["docs_found_in"] = None
+        result["docs_found"] = True
+        assert docs_folder.is_relative_to(outdir)
+        result["docs_folder"] = str(docs_folder.relative_to(outdir))
+        result["md_count"] = int(md_count)
+        if md_count < 10:
+            append_line_threadsafe(
+                md_failed_path, f"{owner}/{repo}  # md-count={md_count}\n", lock
+            )
+            result["status"] = "low-md-count"
+        return result
+
     if prefer_sparse:
         # Try to fetch only docs/ folder via git sparse-checkout without REST API
         for branch_guess in ("main", "master"):
@@ -667,7 +699,7 @@ def process_repo_entry(
                 f"https://github.com/{owner}/{repo}/tree/{default_branch}/{docs_path}"
             )
         elif isinstance(contents, dict) and contents.get("type") == "file":
-            logger.info(f"Found file at docs (single-file). Downloading...")
+            logger.info("Found file at docs (single-file). Downloading...")
             if not dry_run:
                 download_folder_via_api(
                     owner, repo, docs_path, default_branch, saved_root, only_md=only_md
@@ -781,7 +813,8 @@ def _init_duckdb(con):
           size BIGINT,
           mtime BIGINT,
           lang TEXT,
-          content TEXT
+          content TEXT,
+          updated_at TEXT
         );
         """
     )
@@ -801,9 +834,9 @@ def main():
     )
     args = parser.parse_args()
 
-    # Load YAML config next to this script (parse_gh_docs_config.yaml) if present
+    # Load YAML config next to this script (scrape_gh_docs_config.yaml) if present
     cfg: Dict[str, Any] = {}
-    cfg_path = Path(__file__).with_name("parse_gh_docs_config.yaml")
+    cfg_path = Path(__file__).with_name("scrape_gh_docs_config.yaml")
     if cfg_path.exists():
         cfg = yaml.safe_load(cfg_path.read_text(encoding="utf-8")) or {}
 
@@ -827,6 +860,7 @@ def main():
     prefer_zip_value = bool(cfg.get("prefer_zip", False))
     prefer_sparse_value = bool(cfg.get("prefer_sparse", False))
     only_md_value = bool(cfg.get("only_md", False))
+    resume_value = bool(cfg.get("resume", True))
     quiet_value = bool(cfg.get("quiet", False))
     # CLI should override YAML for convenience
     no_fetch_value = bool(args.no_fetch or cfg.get("no_fetch", False))
@@ -869,7 +903,7 @@ def main():
         lines: List[str] = []
         if not input_parquet_values:
             logger.error(
-                "'input_parquet' is required. Configure one or more Parquet files with a 'link' column in parse_gh_docs_config.yaml."
+                "'input_parquet' is required. Configure one or more Parquet files with a 'link' column in scrape_gh_docs_config.yaml."
             )
             sys.exit(2)
         # Read repositories from one or more Parquet files; use 'link' column
@@ -893,6 +927,7 @@ def main():
     duckdb_lock = threading.Lock()
 
     # Process repositories concurrently
+    run_ts = datetime.utcnow().isoformat()
     with tqdm(total=len(lines), desc="Repos") as pbar:
 
         def _run(lr: str):
@@ -906,6 +941,7 @@ def main():
                     prefer_zip=prefer_zip_value,
                     prefer_sparse=prefer_sparse_value,
                     only_md=only_md_value,
+                    resume=resume_value,
                 )
                 if res is not None:
                     with results_lock:
@@ -921,6 +957,7 @@ def main():
                                 outdir,
                                 lang_filter_value,
                                 min_text_chars_value,
+                                run_ts,
                             )
                             if rows_one:
                                 cols = [
@@ -933,6 +970,7 @@ def main():
                                     "mtime",
                                     "lang",
                                     "content",
+                                    "updated_at",
                                 ]
                                 df_one = pd.DataFrame(rows_one, columns=cols)
                                 with duckdb_lock:
@@ -995,6 +1033,7 @@ def main():
         "mtime",
         "lang",
         "content",
+        "updated_at",
     ]
     total_inserted = 0
     with duckdb_lock:
@@ -1010,6 +1049,7 @@ def main():
                     outdir,
                     lang_filter_value,
                     min_text_chars_value,
+                    run_ts,
                 )
                 for d in repo_dirs
             ]

data_collection_utils/{parse_gh_docs_config.yaml → scrape_gh_docs_config.yaml}

@@ -7,12 +7,12 @@
 #  - data_collection_utils/awesome_final_repos.py -> awesome-repos.parquet
 #  - data_collection_utils/top_1000_repos.py -> top-1000-repos.parquet
 input_parquet:
-  - ../output/links_filtered.parquet
+  - ../output/links.filtered.parquet
 
 # Output directories/files
 outdir: ../output/raw_docs
 md_failed: ../md-failed.txt
-texts_parquet: ../output/texts.parquet
+texts_parquet: ../output/cleaned_texts_on_metadata_only.parquet
 
 # Concurrency and behavior
 workers: 1
@@ -22,6 +22,9 @@ quiet: false
 # How often to checkpoint partial outputs (in processed repos)
 checkpoint_every: 50
 
+# Resume: skip refetching repos that already exist under outdir
+resume: true
+
 # Auth
 # Secrets are NOT configured here. Put your GitHub token in a .env file (recommended)
 # or export it in your shell environment. Required env var:

data_collection_utils/top-1000-repos.parquet

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:4fa7b4ed3bbeca1048ed6fbe9e2ccb212043c211785fff53564230b4c5cad876
-size 90891
+oid sha256:c2afb145f66ea26cb5d0392b4807f43bdd8c0a69efa56087ada14799802ecc1d
+size 156344