| from __future__ import annotations |
|
|
| import sys |
| from pathlib import Path |
|
|
| ROOT_DIR = Path(__file__).resolve().parents[1] |
| if str(ROOT_DIR) not in sys.path: |
| sys.path.insert(0, str(ROOT_DIR)) |
|
|
| import pandas as pd |
|
|
| from pipeline_00_config import ( |
| ENGLISH_COMMENTS_CSV, |
| ENGLISH_COMMENTS_PERIOD_CSV, |
| LANGUAGE_FILTER_COLUMNS, |
| LEGACY_TOPIC_COMMENTS_CSV, |
| LEGACY_TOPIC_DESCRIPTIONS_CSV, |
| TOPIC_ALIGNED_COMMENTS_CSV, |
| TOPIC_BASE_COMMENTS_CSV, |
| TOPIC_COVERAGE_REPORT_TXT, |
| TOPIC_DESCRIPTIONS_CSV, |
| ensure_directories, |
| load_language_comment_file, |
| normalize_series, |
| read_csv_with_fallback, |
| ) |
|
|
|
|
| def main() -> None: |
| ensure_directories() |
|
|
| comments_full, repaired_full = load_language_comment_file(ENGLISH_COMMENTS_CSV, LANGUAGE_FILTER_COLUMNS) |
| comments_period, repaired_period = load_language_comment_file(ENGLISH_COMMENTS_PERIOD_CSV, LANGUAGE_FILTER_COLUMNS) |
| if repaired_full: |
| comments_full.to_csv(ENGLISH_COMMENTS_CSV, index=False) |
| if repaired_period: |
| comments_period.to_csv(ENGLISH_COMMENTS_PERIOD_CSV, index=False) |
|
|
| base = read_csv_with_fallback(TOPIC_BASE_COMMENTS_CSV, low_memory=False) |
| legacy_comments = read_csv_with_fallback(LEGACY_TOPIC_COMMENTS_CSV, sep=";", engine="python") |
| legacy_descriptions = read_csv_with_fallback(LEGACY_TOPIC_DESCRIPTIONS_CSV, sep=";", engine="python") |
|
|
| comments_period = comments_period.copy() |
| comments_period["comment_normalized"] = normalize_series(comments_period["comment"]) |
| comments_period["comment_length"] = comments_period["comment_normalized"].str.len() |
| blank_removed = int((comments_period["comment_normalized"] == "").sum()) |
| short_removed = int(((comments_period["comment_normalized"] != "") & (comments_period["comment_length"] < 15)).sum()) |
| duplicate_removed = int( |
| comments_period.loc[comments_period["comment_length"] >= 15, "comment_normalized"].duplicated().sum() |
| ) |
|
|
| legacy_comments = legacy_comments.copy() |
| legacy_comments["comment_normalized"] = normalize_series(legacy_comments["comment"]) |
| legacy_comments["topic_id"] = pd.to_numeric(legacy_comments["topic_id"], errors="coerce") |
| conflict_count = int((legacy_comments.groupby("comment_normalized")["topic_id"].nunique() > 1).sum()) |
| legacy_comments = legacy_comments.sort_values(["comment_normalized", "comment_id"]).drop_duplicates( |
| subset=["comment_normalized"], keep="first" |
| ) |
|
|
| aligned = base.merge( |
| legacy_comments[["comment_normalized", "topic_id", "Name", "Representation"]], |
| on="comment_normalized", |
| how="left", |
| ) |
| aligned = aligned.rename( |
| columns={ |
| "Name": "topic_name", |
| "Representation": "topic_representation", |
| } |
| ) |
| aligned.to_csv(TOPIC_ALIGNED_COMMENTS_CSV, index=False) |
|
|
| descriptions = legacy_descriptions.rename( |
| columns={ |
| "Topic": "topic_id", |
| "Count": "topic_count", |
| "Name": "topic_name", |
| "Representation": "topic_representation", |
| "Representative_Docs": "representative_docs", |
| } |
| ) |
| descriptions["topic_id"] = pd.to_numeric(descriptions["topic_id"], errors="coerce") |
| descriptions = descriptions.dropna(subset=["topic_id"]).drop_duplicates(subset=["topic_id"]).sort_values("topic_id") |
| descriptions.to_csv(TOPIC_DESCRIPTIONS_CSV, index=False) |
|
|
| matched = int(aligned["topic_id"].notna().sum()) |
| unmatched = int(aligned["topic_id"].isna().sum()) |
|
|
| lines = [ |
| "Relatorio de cobertura da etapa de topicos", |
| "=" * 72, |
| f"Comentarios no corpus final em ingles: {len(comments_full)}", |
| f"Comentarios no corpus do periodo analitico: {len(comments_period)}", |
| f"Comentarios vazios removidos: {blank_removed}", |
| f"Comentarios curtos removidos (<15 caracteres): {short_removed}", |
| f"Comentarios duplicados por texto removidos: {duplicate_removed}", |
| f"Comentarios na base de topicos: {len(base)}", |
| f"Comentarios unicos no arquivo legado de topicos: {legacy_comments['comment_normalized'].nunique()}", |
| f"Textos com conflito de topic_id no legado: {conflict_count}", |
| f"Comentarios alinhados com topic_id: {matched}", |
| f"Comentarios sem correspondencia no legado: {unmatched}", |
| "", |
| "Observacao: a base de topicos usa texto normalizado, comentario minimo de 15 caracteres e deduplicacao por texto.", |
| ] |
| TOPIC_COVERAGE_REPORT_TXT.write_text("\n".join(lines) + "\n", encoding="utf-8") |
|
|
| print(f"Base de topicos: {len(base)}") |
| print(f"Alinhados com topicos: {matched}") |
| print(f"Sem correspondencia: {unmatched}") |
| print(f"Relatorio salvo em: {TOPIC_COVERAGE_REPORT_TXT}") |
|
|
|
|
| if __name__ == "__main__": |
| main() |
|
|
