Merge pull request #48 from kudanai/kudanai/cleanup

Move transliterator to it's own package
kudanai · Dec 24, 2023 · e6d3bb8 · e6d3bb8
2 parents ebdf671 + 6b87164
commit e6d3bb8
Show file tree

Hide file tree

Showing 10 changed files with 20 additions and 155 deletions.
diff --git a/utils/__init__.py b/utils/__init__.py
diff --git a/utils/output/repeated_lines.txt b/utils/output/repeated_lines.txt
diff --git a/utils/transliterator/README.md b/utils/transliterator/README.md
@@ -0,0 +1,10 @@
+# Transliterator
+
+This utility aids the generation of the alternate (transliterated) translation file
+using a custom dictionary swap with a few rules applied.
+
+## Usage
+
+```shell
+python ./utils/transliterator/generate_alt_text_with_thikithaana.py
+```
diff --git a/utils/transliterator/__init__.py b/utils/transliterator/__init__.py
diff --git a/utils/generate_alt_text_with_thikithaana.py → ...tor/generate_alt_text_with_thikithaana.py b/utils/generate_alt_text_with_thikithaana.py → ...tor/generate_alt_text_with_thikithaana.py
@@ -4,8 +4,9 @@
 import csv
 import transliterate_arabic_to_thaana as trans
 
-INPUT_FILE = "master_dv.divehi.txt"
-OUTPUT_FILE = "alt_dv.divehi.txt"
+
+INPUT_FILE = trans.BASE_DIR/"master_dv.divehi.txt"
+OUTPUT_FILE = trans.BASE_DIR/"alt_dv.divehi.txt"
 
 
 def main():

diff --git a/utils/transliterate_arabic_to_thaana.py → ...terator/transliterate_arabic_to_thaana.py b/utils/transliterate_arabic_to_thaana.py → ...terator/transliterate_arabic_to_thaana.py
@@ -2,16 +2,19 @@
 
 import re
 import csv
+from pathlib import Path
 
 ARAB_SET = u"\u0621-\u06FF"
 THAA_SET = u"\u0780-\u07B1"
 LONG_FILI = u"\u07A7\u07A9\u07AB\u07AD\u07AF"
 
-INPUT_FILE = "utils/transliteration_test/arabic_words_with_next.txt"
-OUTPUT_FILE = "utils/transliteration_test/arabic_words_with_next_out.csv"
+BASE_DIR = Path(__file__).resolve().parent.parent.parent
 
-TRANSLIT_FILE = "utils/transliteration_data/_quran_freq_dv.csv"
-NEXT_WORDS_FILE = "utils/transliteration_data/_dhivehi_next_words.csv"
+INPUT_FILE = BASE_DIR/"utils/transliterator/transliteration_test/arabic_words_with_next.txt"
+OUTPUT_FILE = BASE_DIR/"utils/transliterator/transliteration_test/arabic_words_with_next_out.csv"
+
+TRANSLIT_FILE = BASE_DIR/"utils/transliterator/transliteration_data/_quran_freq_dv.csv"
+NEXT_WORDS_FILE = BASE_DIR/"utils/transliterator/transliteration_data/_dhivehi_next_words.csv"
 
 tranliteration_dicts = []
 with open(TRANSLIT_FILE, mode='r', encoding='utf-8-sig', newline='') as translit_file:

diff --git a/...nsliteration_data/_dhivehi_next_words.csv → ...nsliteration_data/_dhivehi_next_words.csv b/...nsliteration_data/_dhivehi_next_words.csv → ...nsliteration_data/_dhivehi_next_words.csv
diff --git a/...s/transliteration_data/_quran_freq_dv.csv → ...r/transliteration_data/_quran_freq_dv.csv b/...s/transliteration_data/_quran_freq_dv.csv → ...r/transliteration_data/_quran_freq_dv.csv
diff --git a/...iteration_test/arabic_words_with_next.txt → ...iteration_test/arabic_words_with_next.txt b/...iteration_test/arabic_words_with_next.txt → ...iteration_test/arabic_words_with_next.txt
diff --git a/...ation_test/arabic_words_with_next_out.csv → ...ation_test/arabic_words_with_next_out.csv b/...ation_test/arabic_words_with_next_out.csv → ...ation_test/arabic_words_with_next_out.csv