exaland · December 5, 2025 00:27
diff --git a/extract_strings.py b/extract_strings.py
 #!/usr/bin/env python3
 """
 Simple extractor to find hardcoded strings in Android XML files and move them to res/values/strings.xml.
 - Scans app/src/main/res for layout/menu/xml/etc (skips values/)
 - Targets attributes: android:text, android:hint, android:contentDescription, android:title, android:label, android:summary
 - Skips values already referencing @, ? or data-binding @{...}
 - Adds entries to app/src/main/res/values/strings.xml and replaces literals with @string/<key>
 - Creates locale placeholder values-<lang>/strings.xml for existing locales or a default set

 Usage:
  tools/extract_strings.py [--dry-run]

 Note: This is a best-effort tool. Review changes before committing.
 """

 import os
 import re
 import sys
 import argparse
 import hashlib
 import datetime
 from collections import defaultdict
 from xml.etree import ElementTree as ET

 ROOT = os.path.abspath(os.path.join(os.path.dirname(__file__), '..'))
 APP_RES = os.path.join(ROOT, 'app', 'src', 'main', 'res')
 TARGET_ATTRS = ['text', 'hint', 'contentDescription', 'title', 'label', 'summary']
 ATTR_RE = re.compile(r'android:({})="([^"]*)"'.format('|'.join(TARGET_ATTRS)))
 SKIP_PREFIX = ('@', '?', '@{')


 def load_existing_keys(strings_xml_path):
    keys = {}
    if not os.path.exists(strings_xml_path):
        return keys
    try:
        tree = ET.parse(strings_xml_path)
        root = tree.getroot()
        for child in root.findall('string'):
            name = child.get('name')
            text = ''.join(child.itertext()) if child is not None else ''
            keys[name] = text
    except ET.ParseError:
        # fallback: naive regex parse
        with open(strings_xml_path, 'r', encoding='utf-8') as f:
            data = f.read()
        for m in re.finditer(r'<string\s+name="([^"]+)">(.+?)</string>', data, re.S):
            keys[m.group(1)] = m.group(2)
    return keys


 def ensure_strings_xml(path):
    if not os.path.exists(path):
        os.makedirs(os.path.dirname(path), exist_ok=True)
        with open(path, 'w', encoding='utf-8') as f:
            f.write('<?xml version="1.0" encoding="utf-8"?>\n<resources>\n</resources>\n')


 def make_key(filename, attr, text, existing_keys):
    base = re.sub(r'[^0-9a-zA-Z]+', '_', filename).strip('_').lower()
    snippet = text.strip()
    # Short hash to reduce collisions
    h = hashlib.sha1(snippet.encode('utf-8')).hexdigest()[:8]
    key = f"str_{base}_{attr}_{h}"
    # ensure unique vs existing keys
    i = 1
    orig = key
    while key in existing_keys:
        key = f"{orig}_{i}"
        i += 1
    return key


 def find_locale_dirs(res_dir):
    dirs = []
    for name in os.listdir(res_dir):
        if name.startswith('values-') and os.path.isdir(os.path.join(res_dir, name)):
            dirs.append(name)
    return sorted(dirs)


 def create_locale_placeholders(res_dir, new_entries, locales):
    created = []
    for loc in locales:
        path = os.path.join(res_dir, loc, 'strings.xml')
        ensure_strings_xml(path)
        # naive append: only add if key missing
        keys = load_existing_keys(path)
        if not new_entries:
            continue
        with open(path, 'r', encoding='utf-8') as f:
            data = f.read()
        insert_pos = data.rfind('</resources>')
        if insert_pos == -1:
            continue
        additions = '\n  <!-- TODO: auto-generated placeholders -->\n'
        for k, v in new_entries.items():
            if k in keys:
                continue
            # escape ampersand and others
            v_esc = v.replace('&', '&amp;').replace('<', '&lt;').replace('>', '&gt;')
            additions += f'  <string name="{k}">{v_esc}</string>\n'
        if additions.strip():
            newdata = data[:insert_pos] + additions + data[insert_pos:]
            with open(path, 'w', encoding='utf-8') as f:
                f.write(newdata)
            created.append(path)
    return created


 def main():
    parser = argparse.ArgumentParser()
    parser.add_argument('--dry-run', action='store_true', help='Show planned changes')
    parser.add_argument('--apply', action='store_true', help='Apply changes')
    args = parser.parse_args()

    if not os.path.isdir(APP_RES):
        print('ERROR: res directory not found at', APP_RES)
        sys.exit(1)

    strings_xml = os.path.join(APP_RES, 'values', 'strings.xml')
    ensure_strings_xml(strings_xml)
    existing_keys = load_existing_keys(strings_xml)

    # collect files to scan
    xml_files = []
    for root, dirs, files in os.walk(APP_RES):
        # skip values folders
        if os.path.basename(root).startswith('values'):
            continue
        for fn in files:
            if fn.endswith('.xml'):
                xml_files.append(os.path.join(root, fn))

    planned_changes = []
    new_entries = {}
    files_modified = {}

    for path in xml_files:
        with open(path, 'r', encoding='utf-8') as f:
            text = f.read()
        # skip if no android: attributes present quickly
        if 'android:' not in text:
            continue
        matches = list(ATTR_RE.finditer(text))
        if not matches:
            continue
        repl_text = text
        local_changes = []
        for m in matches:
            attr = m.group(1)
            val = m.group(2)
            if not val or val.startswith(SKIP_PREFIX):
                continue
            # skip if contains data-binding or expressions
            if '@{' in val or '}' in val and val.strip().startswith('{'):
                continue
            # skip if translatable="false" on the tag (simple heuristic)
            # If thestring is mostly numbers or punctuation, skip
            if re.fullmatch(r'[\d\s\W]+', val):
                continue
            filename = os.path.splitext(os.path.basename(path))[0]
            key = make_key(filename, attr, val, existing_keys)
            existing_keys[key] = val
            new_entries[key] = val
            # prepare replacement: escape key reference
            # replace only this exact attribute occurrence
            old_attr = f'android:{attr}="{val}"'
            new_attr = f'android:{attr}="@string/{key}"'
            if old_attr in repl_text:
                repl_text = repl_text.replace(old_attr, new_attr, 1)
                local_changes.append((attr, val, key))
        if local_changes:
            planned_changes.append((path, local_changes))
            files_modified[path] = repl_text

    # show report
    ts = datetime.datetime.now().strftime('%Y%m%d%H%M%S')
    backup_dir = os.path.join(ROOT, 'backup', 'extracted', ts)
    print('Found', len(planned_changes), 'files with hardcoded strings.')
    total = sum(len(v) for _, v in planned_changes)
    print('Planned new string entries:', total)
    for p, changes in planned_changes:
        print('-', p)
        for attr, val, key in changes:
            print(f'   {attr}: "{val}" -> @string/{key}')

    # list locales
    locales = find_locale_dirs(APP_RES)
    if not locales:
        # create default locales
        locales = ['values-fr', 'values-es', 'values-ar', 'values-sv']
        print('No locale directories found; will create defaults:', locales)
    else:
        print('Found locale directories:', locales)

    if args.dry_run:
        print('\nDry run only. No files modified.')
        return

    # apply changes
    os.makedirs(backup_dir, exist_ok=True)
    for path, newtext in files_modified.items():
        rel = os.path.relpath(path, ROOT)
        bak = os.path.join(backup_dir, rel.replace(os.sep, '_'))
        with open(bak, 'w', encoding='utf-8') as f:
            with open(path, 'r', encoding='utf-8') as orig:
                f.write(orig.read())
        with open(path, 'w', encoding='utf-8') as f:
            f.write(newtext)
        print('Modified', path, '(backup at', bak, ')')

    # append new entries to main strings.xml
    if new_entries:
        with open(strings_xml, 'r', encoding='utf-8') as f:
            data = f.read()
        insert_pos = data.rfind('</resources>')
        additions = '\n  <!-- auto-extracted strings - {} -->\n'.format(ts)
        for k, v in new_entries.items():
            v_esc = v.replace('&', '&amp;').replace('<', '&lt;').replace('>', '&gt;')
            additions += f'  <string name="{k}">{v_esc}</string>\n'
        newdata = data[:insert_pos] + additions + data[insert_pos:]
        with open(strings_xml, 'w', encoding='utf-8') as f:
            f.write(newdata)
        print('Updated', strings_xml)

    # create/update locale placeholders
    created = create_locale_placeholders(APP_RES, new_entries, locales)
    for c in created:
        print('Updated locale file', c)

    # write summary
    report_path = os.path.join(ROOT, 'tools', f'extract_strings_report_{ts}.txt')
    with open(report_path, 'w', encoding='utf-8') as rpt:
        rpt.write(f'Extracted at {ts}\n')
        rpt.write(f'Files modified: {len(planned_changes)}\n')
        rpt.write(f'New entries: {len(new_entries)}\n')
        for p, changes in planned_changes:
            rpt.write(p + '\n')
            for attr, val, key in changes:
                rpt.write(f'  {attr}: "{val}" -> {key}\n')
    print('Wrote report to', report_path)


 if __name__ == '__main__':
    main()
	#!/usr/bin/env python3
	"""
	Simple extractor to find hardcoded strings in Android XML files and move them to res/values/strings.xml.
	- Scans app/src/main/res for layout/menu/xml/etc (skips values/)
	- Targets attributes: android:text, android:hint, android:contentDescription, android:title, android:label, android:summary
	- Skips values already referencing @, ? or data-binding @{...}
	- Adds entries to app/src/main/res/values/strings.xml and replaces literals with @string/<key>
	- Creates locale placeholder values-<lang>/strings.xml for existing locales or a default set

	Usage:
	tools/extract_strings.py [--dry-run]

	Note: This is a best-effort tool. Review changes before committing.
	"""

	import os
	import re
	import sys
	import argparse
	import hashlib
	import datetime
	from collections import defaultdict
	from xml.etree import ElementTree as ET

	ROOT = os.path.abspath(os.path.join(os.path.dirname(__file__), '..'))
	APP_RES = os.path.join(ROOT, 'app', 'src', 'main', 'res')
	TARGET_ATTRS = ['text', 'hint', 'contentDescription', 'title', 'label', 'summary']
	ATTR_RE = re.compile(r'android:({})="([^"]*)"'.format('\|'.join(TARGET_ATTRS)))
	SKIP_PREFIX = ('@', '?', '@{')


	def load_existing_keys(strings_xml_path):
	keys = {}
	if not os.path.exists(strings_xml_path):
	return keys
	try:
	tree = ET.parse(strings_xml_path)
	root = tree.getroot()
	for child in root.findall('string'):
	name = child.get('name')
	text = ''.join(child.itertext()) if child is not None else ''
	keys[name] = text
	except ET.ParseError:
	# fallback: naive regex parse
	with open(strings_xml_path, 'r', encoding='utf-8') as f:
	data = f.read()
	for m in re.finditer(r'<string\s+name="([^"]+)">(.+?)</string>', data, re.S):
	keys[m.group(1)] = m.group(2)
	return keys


	def ensure_strings_xml(path):
	if not os.path.exists(path):
	os.makedirs(os.path.dirname(path), exist_ok=True)
	with open(path, 'w', encoding='utf-8') as f:
	f.write('<?xml version="1.0" encoding="utf-8"?>\n<resources>\n</resources>\n')


	def make_key(filename, attr, text, existing_keys):
	base = re.sub(r'[^0-9a-zA-Z]+', '_', filename).strip('_').lower()
	snippet = text.strip()
	# Short hash to reduce collisions
	h = hashlib.sha1(snippet.encode('utf-8')).hexdigest()[:8]
	key = f"str_{base}_{attr}_{h}"
	# ensure unique vs existing keys
	i = 1
	orig = key
	while key in existing_keys:
	key = f"{orig}_{i}"
	i += 1
	return key


	def find_locale_dirs(res_dir):
	dirs = []
	for name in os.listdir(res_dir):
	if name.startswith('values-') and os.path.isdir(os.path.join(res_dir, name)):
	dirs.append(name)
	return sorted(dirs)


	def create_locale_placeholders(res_dir, new_entries, locales):
	created = []
	for loc in locales:
	path = os.path.join(res_dir, loc, 'strings.xml')
	ensure_strings_xml(path)
	# naive append: only add if key missing
	keys = load_existing_keys(path)
	if not new_entries:
	continue
	with open(path, 'r', encoding='utf-8') as f:
	data = f.read()
	insert_pos = data.rfind('</resources>')
	if insert_pos == -1:
	continue
	additions = '\n <!-- TODO: auto-generated placeholders -->\n'
	for k, v in new_entries.items():
	if k in keys:
	continue
	# escape ampersand and others
	v_esc = v.replace('&', '&').replace('<', '<').replace('>', '>')
	additions += f' <string name="{k}">{v_esc}</string>\n'
	if additions.strip():
	newdata = data[:insert_pos] + additions + data[insert_pos:]
	with open(path, 'w', encoding='utf-8') as f:
	f.write(newdata)
	created.append(path)
	return created


	def main():
	parser = argparse.ArgumentParser()
	parser.add_argument('--dry-run', action='store_true', help='Show planned changes')
	parser.add_argument('--apply', action='store_true', help='Apply changes')
	args = parser.parse_args()

	if not os.path.isdir(APP_RES):
	print('ERROR: res directory not found at', APP_RES)
	sys.exit(1)

	strings_xml = os.path.join(APP_RES, 'values', 'strings.xml')
	ensure_strings_xml(strings_xml)
	existing_keys = load_existing_keys(strings_xml)

	# collect files to scan
	xml_files = []
	for root, dirs, files in os.walk(APP_RES):
	# skip values folders
	if os.path.basename(root).startswith('values'):
	continue
	for fn in files:
	if fn.endswith('.xml'):
	xml_files.append(os.path.join(root, fn))

	planned_changes = []
	new_entries = {}
	files_modified = {}

	for path in xml_files:
	with open(path, 'r', encoding='utf-8') as f:
	text = f.read()
	# skip if no android: attributes present quickly
	if 'android:' not in text:
	continue
	matches = list(ATTR_RE.finditer(text))
	if not matches:
	continue
	repl_text = text
	local_changes = []
	for m in matches:
	attr = m.group(1)
	val = m.group(2)
	if not val or val.startswith(SKIP_PREFIX):
	continue
	# skip if contains data-binding or expressions
	if '@{' in val or '}' in val and val.strip().startswith('{'):
	continue
	# skip if translatable="false" on the tag (simple heuristic)
	# If thestring is mostly numbers or punctuation, skip
	if re.fullmatch(r'[\d\s\W]+', val):
	continue
	filename = os.path.splitext(os.path.basename(path))[0]
	key = make_key(filename, attr, val, existing_keys)
	existing_keys[key] = val
	new_entries[key] = val
	# prepare replacement: escape key reference
	# replace only this exact attribute occurrence
	old_attr = f'android:{attr}="{val}"'
	new_attr = f'android:{attr}="@string/{key}"'
	if old_attr in repl_text:
	repl_text = repl_text.replace(old_attr, new_attr, 1)
	local_changes.append((attr, val, key))
	if local_changes:
	planned_changes.append((path, local_changes))
	files_modified[path] = repl_text

	# show report
	ts = datetime.datetime.now().strftime('%Y%m%d%H%M%S')
	backup_dir = os.path.join(ROOT, 'backup', 'extracted', ts)
	print('Found', len(planned_changes), 'files with hardcoded strings.')
	total = sum(len(v) for _, v in planned_changes)
	print('Planned new string entries:', total)
	for p, changes in planned_changes:
	print('-', p)
	for attr, val, key in changes:
	print(f' {attr}: "{val}" -> @string/{key}')

	# list locales
	locales = find_locale_dirs(APP_RES)
	if not locales:
	# create default locales
	locales = ['values-fr', 'values-es', 'values-ar', 'values-sv']
	print('No locale directories found; will create defaults:', locales)
	else:
	print('Found locale directories:', locales)

	if args.dry_run:
	print('\nDry run only. No files modified.')
	return

	# apply changes
	os.makedirs(backup_dir, exist_ok=True)
	for path, newtext in files_modified.items():
	rel = os.path.relpath(path, ROOT)
	bak = os.path.join(backup_dir, rel.replace(os.sep, '_'))
	with open(bak, 'w', encoding='utf-8') as f:
	with open(path, 'r', encoding='utf-8') as orig:
	f.write(orig.read())
	with open(path, 'w', encoding='utf-8') as f:
	f.write(newtext)
	print('Modified', path, '(backup at', bak, ')')

	# append new entries to main strings.xml
	if new_entries:
	with open(strings_xml, 'r', encoding='utf-8') as f:
	data = f.read()
	insert_pos = data.rfind('</resources>')
	additions = '\n <!-- auto-extracted strings - {} -->\n'.format(ts)
	for k, v in new_entries.items():
	v_esc = v.replace('&', '&').replace('<', '<').replace('>', '>')
	additions += f' <string name="{k}">{v_esc}</string>\n'
	newdata = data[:insert_pos] + additions + data[insert_pos:]
	with open(strings_xml, 'w', encoding='utf-8') as f:
	f.write(newdata)
	print('Updated', strings_xml)

	# create/update locale placeholders
	created = create_locale_placeholders(APP_RES, new_entries, locales)
	for c in created:
	print('Updated locale file', c)

	# write summary
	report_path = os.path.join(ROOT, 'tools', f'extract_strings_report_{ts}.txt')
	with open(report_path, 'w', encoding='utf-8') as rpt:
	rpt.write(f'Extracted at {ts}\n')
	rpt.write(f'Files modified: {len(planned_changes)}\n')
	rpt.write(f'New entries: {len(new_entries)}\n')
	for p, changes in planned_changes:
	rpt.write(p + '\n')
	for attr, val, key in changes:
	rpt.write(f' {attr}: "{val}" -> {key}\n')
	print('Wrote report to', report_path)


	if __name__ == '__main__':
	main()
No results found