From ee8f3f60816d7d47056616df2a22be1049e812fc Mon Sep 17 00:00:00 2001
From: "Antonio J. Delgado" <ad@susurrando.com>
Date: Thu, 24 Jul 2025 12:31:27 +0300
Subject: [PATCH] Add initial script

---
 extract_words/extract_words.py | 47 ++++++++++++++++++++++++++++++++++
 1 file changed, 47 insertions(+)

diff --git a/extract_words/extract_words.py b/extract_words/extract_words.py
index 8ada551..e99a15f 100755
--- a/extract_words/extract_words.py
+++ b/extract_words/extract_words.py
@@ -9,6 +9,7 @@ import sys
 import os
 import json
 import time
+import re
 import logging
 from logging.handlers import SysLogHandler
 import click
@@ -48,6 +49,38 @@ class ExtractWords:
         }
         self.data = self._read_cached_data()
 
+        self._debug(
+            f"Reading file '{self.config['input_file'].name}'..."
+        )
+        content = self.config['input_file'].read()
+        self._debug(
+            "Unsplitting separated words..."
+        )
+        content = re.sub('- ?\n', '', content)
+        content = content.lower()
+        self._debug(
+            f"Splitting file of {len(content)} bytes..."
+        )
+        words = content.split()
+        self._debug(
+            f"Found {len(words)} a total (non-unique) words"
+        )
+        all_words = []
+        for word in words:
+            match = re.match(r'\w*', word)
+            if match:
+                if match.group(0) != '':
+                    match_numbers = re.match(r'[0-9]', match.group(0))
+                    if not match_numbers:
+                        if match.group(0) not in all_words:
+                            all_words.append(match.group(0))
+        self._debug(
+            f"A total of {len(all_words)} unique words"
+        )
+        all_words.sort()
+        self.config['output_file'].write('\n'.join(all_words))
+
+
     def close(self):
         '''Close class and save data'''
         self._save_cached_data(self.data)
@@ -181,6 +214,20 @@ class ExtractWords:
     default=60*60*24*7,
     help='Max age in seconds for the cache'
 )
+@click.option(
+    '--input-file',
+    '-i',
+    required=True,
+    type=click.File('r'),
+    help='File containing words mixed with other things (like a book) in plain text',
+)
+@click.option(
+    '--output-file',
+    '-O',
+    required=True,
+    type=click.File('w'),
+    help='File to write the words',
+)
 # @click.option("--dummy","-n", is_flag=True,
 # help="Don't do anything, just show what would be done.")
 @click_config_file.configuration_option()