gadict: comparison py/gadict_srs

equal deleted inserted replaced

-:6d4a074cea27
+:2d488cfc4c0c
 # -*- coding: utf-8 -*-
+"""Anki card writer"""
 import os
 import io
 import sys
 import codecs
 import shutil
 import signal
 import regex
 import gadict
+import gadict_freq
 FINAME = None
 FONAME = None
 LANGS = None
+FREQ_SOURCES = []
 # -lang:ru,uk
 ARG_LANG_RE = regex.compile("-lang:(.+)")
 # -freq:var:TAG=FILE or -freq:freq:TAG=FILE
 ARG_FREQ_RE = regex.compile("-freq:(freq|var):([^=]+)=(.+)")
 if len(lang) != 2:
 raise Exception("Incorrect language specification: '{:s}'".format(arg))
 continue
 m = ARG_FREQ_RE.match(arg)
 if m:
-LANGS = set(arg.split(","))
+mode = m.group(1)
-for lang in LANGS:
+tag = m.group(2)
-if len(lang) != 2:
+fname = m.group(3)
-raise Exception("Incorrect language specification: '{:s}'".format(arg))
+with io.open(fname, mode='r', buffering=1, encoding="utf-8") as stream:
+if mode == "var":
+parser = gadict_freq.HeadVarParser(stream)
+elif mode == "freq":
+parser = gadict_freq.FreqlistParser(stream)
+else:
+raise Exception("Unsupported mode: '{:s}'".format(mode))
+wlist = parser.parse()
+FREQ_SOURCES.append((tag, set(wlist)))
 continue
 if arg.startswith("-"):
 raise Exception("Unsupported option format: '{:s}'".format(arg))
 if not FINAME:
 FINAME = arg
 div.glos .lang {
 color: brown;
 }
 span.glos {
 font-size: .95em;
+}
+.freq {
+color: red;
+font-weight: bold;
 }
 .del {
 color: red;
 font-weight: bold;
 }
 builder.add_note("singular", identity, warnmsg, warnmsg+" singular", "del")
 builder.add_note("plural", identity, warnmsg, warnmsg+" plural", "del")
 for (headwords, translations) in DOM[1:]:
 identity = headwords[0].headword
+freqtags = []
+for (freqtag, freqset) in FREQ_SOURCES:
+if identity in freqset:
+freqtags.append(freqtag)
+freqmsg = None
+if len(freqtags) > 0:
+freqmsg = ",".join(freqtags)
+freqmsg = "<div class='freq'>{:s}</div>".format(freqmsg)
 buf = []
 v1, v2, v3 = (None, None, None)
 singular, plural = (None, None)
 for hw in headwords:
 buf.append("<div clsas='headword'>")
 if 's' in hw.attrs:
 singular = (hw.headword, hw.pron)
 if 'pl' in hw.attrs:
 plural = (hw.headword, hw.pron)
 buf.append("</div>")
+if freqmsg:
+buf.append(freqmsg)
 direct_from = "".join(buf)
 buf = []
 for sense in translations:
 write_sense(buf, sense, with_examples = True)
 direct_to = "".join(buf)

changeset 646	2d488cfc4c0c
parent 644	e38cd6112193
child 647	6ae5399c8087