From 77323821cb5db6aa08df8d706fa2da8159058e9a Mon Sep 17 00:00:00 2001 From: nathan Date: Mon, 11 May 2026 10:24:32 -0400 Subject: [PATCH] removed highlighting text. Added in magnifying words left side, sync scrolling, keyboard shortcuts, editing text etc --- scripts/extract_positions.py | 16 +++++++++------- scripts/ocr_wardiaries.py | 2 +- 2 files changed, 10 insertions(+), 8 deletions(-) diff --git a/scripts/extract_positions.py b/scripts/extract_positions.py index 910f63a..bdd5510 100644 --- a/scripts/extract_positions.py +++ b/scripts/extract_positions.py @@ -6,9 +6,11 @@ Outputs a JSON array of position objects. import re import json from html.parser import HTMLParser +from pathlib import Path -INPUT_FILE = r"C:\Users\natha\IdeaProjects\AI-Prototype\Inputs\ocr-output\Calgary-Highlanders_War-Diary_Sep44_olmocr.md" -OUTPUT_FILE = r"C:\Users\natha\IdeaProjects\AI-Prototype\outputs\Calgary-Highlanders_Sep44_positions.json" +_PROJECT_ROOT = Path(__file__).resolve().parent.parent +INPUT_FILE = _PROJECT_ROOT / "Inputs" / "ocr-output" / "Calgary-Highlanders_War-Diary_Sep44_olmocr.md" +OUTPUT_FILE = _PROJECT_ROOT / "outputs" / "Calgary-Highlanders_Sep44_positions.json" # ── helpers ────────────────────────────────────────────────────────────────── @@ -671,10 +673,8 @@ def assign_end_of_day(positions: list[dict]) -> list[dict]: # ── run ─────────────────────────────────────────────────────────────────────── if __name__ == "__main__": - import pathlib - print("Reading source file …") - text = pathlib.Path(INPUT_FILE).read_text(encoding="utf-8") + text = INPUT_FILE.read_text(encoding="utf-8") print("Extracting positions …") positions = extract_positions(text) @@ -691,8 +691,8 @@ if __name__ == "__main__": print("Assigning end-of-day flags …") unique = assign_end_of_day(unique) - pathlib.Path(OUTPUT_FILE).parent.mkdir(parents=True, exist_ok=True) - pathlib.Path(OUTPUT_FILE).write_text( + OUTPUT_FILE.parent.mkdir(parents=True, exist_ok=True) + OUTPUT_FILE.write_text( json.dumps(unique, indent=2, ensure_ascii=False), encoding="utf-8" ) @@ -716,6 +716,8 @@ if __name__ == "__main__": + + diff --git a/scripts/ocr_wardiaries.py b/scripts/ocr_wardiaries.py index 92a2d69..4be960e 100644 --- a/scripts/ocr_wardiaries.py +++ b/scripts/ocr_wardiaries.py @@ -10,7 +10,7 @@ Requirements: Poppler must be on PATH (pdftoppm command must work). Usage: - python ocr_wardiaries.py --input_dir "C:\path\to\pdfs" --output_dir "C:\path\to\output" --api_key "YOUR_KEY" + python ocr_wardiaries.py --input_dir "C:/path/to/pdfs" --output_dir "C:/path/to/output" --api_key "YOUR_KEY" """ import argparse