HeliBoard/tools/morekeys_reader.py
Helium314 cbfa934721
Switch most latin layouts to use the simple parser (#276)
And do some tweaks so it works better
2023-11-19 21:16:23 +01:00

245 lines
9.4 KiB
Python

#!/bin/python
import pathlib
import xml.etree.ElementTree as ET
import langcodes
import re
# (WIP) script for reading moreKeys from tools used to create KeyTextsTable while resolving "!text/" references
# plan is to create a straightforward moreKeys file per language
xml_folder = "make-keyboard-text/src/main/resources/"
default_file = pathlib.Path(__file__).parent / f"{xml_folder}/values/donottranslate-more-keys.xml"
out_folder = pathlib.Path(__file__).parent / "../app/src/main/assets/language_key_texts/"
def append_to_morekeys(morekeys, text, label):
if label in morekeys:
morekeys[label] = morekeys[label] + " " + text
else:
morekeys[label] = text
def prepend_to_morekeys(morekeys, text, label):
if label in morekeys:
morekeys[label] = text + " " + morekeys[label]
else:
morekeys[label] = text
def read_keys(file):
root = ET.parse(file).getroot()
morekeys = dict()
extra_keys = dict()
for key in root.iter("resources"):
for string in key.iter("string"):
for tag, value in string.items():
if tag != "name":
print("ignoring tag " + tag)
continue
if string.text is None:
text = ""
else:
text = resolve_text(string.text, file)
if "!text/" in text:
raise ValueError(f"can't have !text/ in {text} (from {string.text})")
if " " in text:
raise ValueError(f"can't have consecutive spaces in {text} (from {string.text})")
if value.startswith("keyspec_") and "_row" in value:
# put additional key labels (for nordic, spanish, swiss)
key = value.split("_row")[1]
d = extra_keys.get(key, dict())
d["label"] = text
extra_keys[key] = d
elif value.startswith("morekeys_") and "_row" in value:
# put additional key morekeys (for nordic, spanish, swiss)
key = value.split("_row")[1]
d = extra_keys.get(key, dict())
d["morekeys"] = text
extra_keys[key] = d
elif value.startswith("morekeys_"):
key_label = value.split("morekeys_")[1]
if len(key_label) > 1 and key_label != "punctuation":
print(f"ignoring long more key: {key_label}: {text}")
continue
morekeys[key_label] = text
elif value == "single_quotes":
prepend_to_morekeys(morekeys, text, '\'')
elif value == "single_angle_quotes":
append_to_morekeys(morekeys, text, '\'')
elif value == "double_quotes":
prepend_to_morekeys(morekeys, text, '\"')
elif value == "double_angle_quotes":
append_to_morekeys(morekeys, text, '\"')
# todo: labels should be in [labels] and use sth like symbols: ?123
else:
print(f"ignored tag: {tag}={value}, {text}")
keys = dict()
keys["morekeys"] = morekeys
keys["extra_keys"] = extra_keys
return keys
def resolve_text(text, file):
if text.startswith("\"") and text.endswith("\"") and len(text) > 1:
text = text[1:-1]
sp = re.split("(?<!\\\\),", text)
if len(sp) > 1: # resolve each entry separately
result = []
for t in sp:
resolved = resolve_text(t, file)
if text.startswith("\\"): # remove backslash at start, this seems to be happening somewhere in android parsing too
result.append(resolved[1:])
else:
result.append(resolved)
return " ".join(result) # join with space, because that doesn't cause issues with comma in moreKeys
if "!text/" not in text:
if text.startswith("\\"): # see above
return text[1:]
return text
root = ET.parse(file).getroot()
sp = text.split("!text/")
required = sp[1]
for key in root.findall(".//string"):
for tag, value in key.items():
if tag == "name" and value == required:
return resolve_text(key.text, file)
# fall back to searching in no-language values
root = ET.parse(default_file).getroot()
for key in root.findall(".//string"):
for tag, value in key.items():
if tag == "name" and value == required:
return resolve_text(key.text, file)
raise LookupError(text + " not found in " + str(file))
def read_locale_from_folder(folder):
if folder.startswith("values-"):
return folder.split("values-")[1]
return None
def write_keys(outfile, keys, locc=""):
with open(outfile, "w") as f:
# write section [more_keys], then [extra_keys], skip if empty
if len(keys["morekeys"]) > 0:
f.write("[morekeys]\n")
for k, v in keys["morekeys"].items():
f.write(f"{k} {v}\n")
if len(keys["extra_keys"]) > 0:
f.write("\n")
if len(keys["extra_keys"]) > 0 and locc != "eo": # eo has the extra key moved into the layout
f.write("[extra_keys]\n")
# clarify somewhere that extra keys only apply to default layout (where to get?)
for k, v in sorted(keys["extra_keys"].items()):
row = k.split("_")[0]
morekeys = v.get("morekeys", "")
label = v["label"]
if len(morekeys) == 0:
f.write(f"{row}: {label}\n")
else:
f.write(f"{row}: {label} {morekeys}\n")
def get_morekeys_texts(write=False):
val = []
for file in (pathlib.Path(__file__).parent / xml_folder).iterdir():
locc = read_locale_from_folder(file.name)
if locc is None:
continue
loc = langcodes.Language.get(locc)
script = loc.assume_script().script
# some scripts are not detected, fill in the current state of OpenBoard
if locc == "sr" or locc == "ky" or locc == "mn":
script = "Cyrl"
if locc == "sr-rZZ" or locc == "uz" or locc == "zz" or locc == "az" or locc == "tl":
script = "Latn"
if script is None:
raise ValueError("undefined script")
if script != "Latn":
continue # skip non-latin scripts for now
print(file)
keys = read_keys(f"{file}/donottranslate-more-keys.xml")
val.append(keys)
if not write:
continue
outfile_name = locc.replace("-r", "_").lower() + ".txt"
outfile = pathlib.Path(out_folder + outfile_name)
outfile.parent.mkdir(exist_ok=True, parents=True)
write_keys(outfile, keys, locc)
return val
def write_combined_lists(keys):
infos_by_letters = dict()
for key in keys:
for k, v in key["morekeys"].items():
infos = infos_by_letters.get(k, dict())
for l in v.split(" "):
if l == "%":
continue
infos[l] = infos.get(l, 0) + 1
infos_by_letters[k] = infos
with open(out_folder + "all_more_keys.txt", 'w') as f:
f.write("[morekeys]\n")
for letter, info in infos_by_letters.items():
f.write(letter + " " + " ".join(info.keys()) + "\n")
with open(out_folder + "more_more_keys.txt", 'w') as f:
f.write("[morekeys]\n")
for letter, info in infos_by_letters.items():
morekeys = []
for morekey, count in info.items():
if count > 1:
morekeys.append(morekey)
if len(morekeys) > 0:
f.write(letter + " " + " ".join(morekeys) + "\n")
def main():
# k = read_keys(default_file)
# write_keys(pathlib.Path(__file__).parent / f"defaultkeys.txt", k)
keys = get_morekeys_texts()
write_combined_lists(keys)
# need to check strings:
# latin, but only in symbol layout
# single_quotes, double_quotes (both used in morekeys of single/double quotes in symbol keyboard)
# single_angle_quotes, double_angle_quotes (same place as above -> merge into the same base ' or ")
# -> just treat them like morekeys_' and morekeys_"
# ... resolving those is really horrible, check different things and maybe include all if not too much?
# latin, but for layout and not for moreKeys
# keyspec_nordic_row (+swiss and spanish) -> normal keys, what do? really specify a layout? or allow modifying?
# keyspec_q + w, y, x (eo only -> hmm, have a separate layout?)
# not latin, but cyrillic (and maybe other non-latin)
# keyspec_east_slavic_row
# keylabel_to_alpha
# label_go_key and other keys (hi-rZZ and sr-rZZ -> why here? they should be in app strings, right?)
# not in latin (so far)
# keyspec_symbols
# additional_morekeys_symbols
# keyspec_currency
# keylabel_to_symbol
# keyspec_comma
# keyhintlabel_period -> that's with the shifted key hint maybe
# keyhintlabel_tablet_period
# keyspec_period
# keyspec_tablet_period
# keyspec_symbols_question
# keyspec_symbols_semicolon
# keyspec_symbols_percent
# keyspec_tablet_comma
# keyhintlabel_tablet_comma
# keyspec_left_parenthesis + right
# keyspec_left_square_bracket + right
# keyspec_left_curly_bracket + right
# keyspec_less_than + greater
# keyspec_less_than_equal + greater
# keyspec_left_double_angle_quote + right
# keyspec_left_single_angle_quote + right
if __name__ == "__main__":
main()