server avoid spurious settings files; update identifier wording

server/gradint.cgi

@@ -2,7 +2,7 @@
 # -*- coding: utf-8 -*-
 #  (either Python 2 or Python 3)
 
-program_name = "gradint.cgi v1.33 (c) 2011,2015,2017-23 Silas S. Brown.  GPL v3+"
+program_name = "gradint.cgi v1.34 (c) 2011,2015,2017-23 Silas S. Brown.  GPL v3+"
 
 #    This program is free software; you can redistribute it and/or modify
 #    it under the terms of the GNU General Public License as published by
@@ -19,7 +19,7 @@ path_add = "$HOME/gradint/bin" # include sox, lame, espeak, maybe oggenc
 lib_path_add = "$HOME/gradint/lib"
 espeak_data_path = "$HOME/gradint"
 
-import os, os.path, sys, cgi, urllib, time
+import os, os.path, sys, cgi, urllib, time, re
 try: from commands import getoutput # Python 2
 except: from subprocess import getoutput # Python 3
 try: from urllib import quote,quote_plus,unquote # Python 2
@@ -74,8 +74,10 @@ reinit_gradint()
 
 def main():
   if "id" in query: # e.g. from redirectHomeKeepCookie
-    os.environ["HTTP_COOKIE"]="id="+query.getfirst("id")
-    print ('Set-Cookie: id=' + query.getfirst("id")+'; expires=Wed, 1 Dec 2036 23:59:59 GMT')
+    queryID = query.getfirst("id")
+    if not re.match("[A-Za-z0-9_.-]",queryID): return htmlOut("Bad query.  Bad, bad query.") # to avoid cluttering the disk if we're being given random queries by an attacker.  IDs we generate are numeric only, but allow alphanumeric in case server admin wants to generate them.  Don't allow =, parens, etc (likely random SQL query)
+    os.environ["HTTP_COOKIE"]="id="+queryID
+    print ('Set-Cookie: id=' + queryID+'; expires=Wed, 1 Dec 2036 23:59:59 GMT') # TODO: S2G
   if has_userID(): setup_userID() # always, even for justSynth, as it may include a voice selection (TODO consequently being called twice in many circumstances, could make this more efficient)
   filetype=""
   if "filetype" in query: filetype=query.getfirst("filetype")
@@ -95,17 +97,17 @@ def main():
     gradint.justSynthesize="0"
     if "l2w" in query and query.getfirst("l2w"):
       gradint.startBrowser=lambda *args:0
-      if query.getfirst("l2")=="zh" and gradint.sanityCheck(query.getfirst("l2w"),"zh"): gradint.justSynthesize += "#en Pinyin needs tones.  Please go back and add tone numbers." # speaking it because alert box might not work and we might be being called from HTML5 Audio stuff (TODO maybe duplicate sanityCheck in js, if so don't call HTML5 audio, then we can have an on-screen message here)
+      if query.getfirst("l2")=="zh" and gradint.generalCheck(query.getfirst("l2w"),"zh"): gradint.justSynthesize += "#en Pinyin needs tones.  Please go back and add tone numbers." # speaking it because alert box might not work and we might be being called from HTML5 Audio stuff (TODO maybe duplicate generalCheck in js, if so don't call HTML5 audio, then we can have an on-screen message here)
       else: gradint.justSynthesize += "#"+query.getfirst("l2").replace("#","").replace('"','')+" "+query.getfirst("l2w").replace("#","").replace('"','')
     if "l1w" in query and query.getfirst("l1w"): gradint.justSynthesize += "#"+query.getfirst("l1").replace("#","").replace('"','')+" "+query.getfirst("l1w").replace("#","").replace('"','')
     if gradint.justSynthesize=="0": return htmlOut('You must type a word in the box before pressing the Speak button.'+backLink) # TODO maybe add a Javascript test to the form also, IF can figure out if window.alert works
     serveAudio(stream = len(gradint.justSynthesize)>100, filetype=filetype)
   elif "add" in query: # add to vocab (l1,l2 the langs, l1w,l2w the words)
     if "l2w" in query and query.getfirst("l2w") and "l1w" in query and query.getfirst("l1w"):
       gradint.startBrowser=lambda *args:0
-      if query.getfirst("l2")=="zh": scmsg=gradint.sanityCheck(query.getfirst("l2w"),"zh")
-      else: scmsg=None
-      if scmsg: htmlOut(gradint.B(scmsg)+gradint.B(backLink))
+      if query.getfirst("l2")=="zh": gcmsg=gradint.generalCheck(query.getfirst("l2w"),"zh")
+      else: gcmsg=None
+      if gcmsg: htmlOut(gradint.B(gcmsg)+gradint.B(backLink))
       else: addWord(query.getfirst("l1w"),query.getfirst("l2w"),query.getfirst("l1"),query.getfirst("l2"))
     else: htmlOut('You must type words in both boxes before pressing the Add button.'+backLink) # TODO maybe add a Javascript test to the form also, IF can figure out a way to tell whether window.alert() works or not
   elif "bulkadd" in query: # bulk adding, from authoring options

src/frontend.py

@@ -1295,10 +1295,10 @@ def openDirectory(dir,inGuiThread=0):
         if inGuiThread: tkMessageBox.showinfo(app.master.title(),msg)
         else: waitOnMessage(msg)
 
-def sanityCheck(text,language,pauseOnError=0): # text is utf-8; returns error message if any
+def generalCheck(text,language,pauseOnError=0): # text is utf-8; returns error message if any
     if not text: return # always OK empty strings
     if pauseOnError:
-        ret = sanityCheck(text,language)
+        ret = generalCheck(text,language)
         if ret: waitOnMessage(ret)
         return ret
     if language=="zh":
@@ -1329,7 +1329,7 @@ def s60_addVocab():
     result = appuifw.multi_query(label1,label2) # unfortunately multi_query can't take default items (and sometimes no T9!), but Form is too awkward (can't see T9 mode + requires 2-button save via Options) and non-multi query would be even more modal
     if not result: return # cancelled
     l2,l1 = result # guaranteed to both be populated
-    while sanityCheck(l2.encode('utf-8'),secondLanguage,1):
+    while generalCheck(l2.encode('utf-8'),secondLanguage,1):
         l2=appuifw.query(label1,"text",u"")
         if not l2: return # cancelled
     # TODO detect duplicates like Tk GUI does?
@@ -1371,7 +1371,7 @@ def s60_viewVocab():
           oldL1,oldL2 = l1,l2
           if action==2:
               first=1
-              while first or (l2 and sanityCheck(l2.encode('utf-8'),secondLanguage,1)):
+              while first or (l2 and generalCheck(l2.encode('utf-8'),secondLanguage,1)):
                   first=0 ; l2=appuifw.query(ensure_unicode(secondLanguage),"text",l2)
               if not l2: continue
           elif action==3:
@@ -1386,7 +1386,7 @@ def s60_viewVocab():
 def android_addVocab():
   while True:
     l2 = None
-    while not l2 or sanityCheck(l2.encode('utf-8'),secondLanguage,1):
+    while not l2 or generalCheck(l2.encode('utf-8'),secondLanguage,1):
       l2 = android.dialogGetInput("Add word","Word in %s" % localise(secondLanguage)).result
       if not l2: return # cancelled
     l1 = android.dialogGetInput("Add word","Meaning in %s" % localise(firstLanguage)).result
@@ -1585,7 +1585,7 @@ def gui_event_loop():
             if not text1 and not text2: app.todo.alert=u"Before pressing the "+localise("Speak")+u" button, you need to type the text you want to hear into the box."
             else:
               if text1.startswith(B('#')): msg="" # see below
-              else: msg=sanityCheck(text1,secondLanguage)
+              else: msg=generalCheck(text1,secondLanguage)
               if msg: app.todo.alert=ensure_unicode(msg)
               else:
                 app.set_watch_cursor = 1 ; app.toRestore = []
@@ -1706,7 +1706,7 @@ def scanDirs():
                 app.todo.alert=msg+" "+localise("Repeat count is 0, so we cannot reduce it for extra revision.")
         elif menu_response=="add":
             text1 = asUnicode(app.Text1.get()).encode('utf-8') ; text2 = asUnicode(app.Text2.get()).encode('utf-8')
-            msg=sanityCheck(text1,secondLanguage)
+            msg=generalCheck(text1,secondLanguage)
             if msg: app.todo.alert=ensure_unicode(msg)
             else:
                 o=appendVocabFileInRightLanguages()

src/synth.py

@@ -1162,7 +1162,7 @@ def abspath_from_start(p): # for just_synthesize to check for paths relative to
     os.chdir(d)
     return r
 
-def just_synthesize(callSanityCheck=0,lastLang_override=None):
+def just_synthesize(callGeneralCheck=0,lastLang_override=None):
     # Handle the justSynthesize setting (see advanced.txt)
     global startAnnouncement,endAnnouncement,logFile,synth_partials_cache
     synth_partials_cache = {} # to stop 'memory leak' when running from the GUI
@@ -1197,7 +1197,7 @@ def checkCanSynth(fname):
                 r = repr(l[0])
                 if r[:1]=="b": r=r[1:]
                 show_warning("Assuming that %s is a word to synthesize in language '%s'" % (r,lastLanguage))
-                if callSanityCheck and sanityCheck(l[0],lastLanguage,1): return
+                if callGeneralCheck and generalCheck(l[0],lastLanguage,1): return
                 event = checkCanSynth("!synth:"+S(l[0])+"_"+S(lastLanguage))
                 if not event: continue # couldn't synth
                 called_synth = 1
@@ -1217,10 +1217,10 @@ def checkCanSynth(fname):
                         lastLanguage=lang ; continue
                     # otherwise, user might have omitted lang by mistake
                     show_warning("Assuming %s was meant to be synthesized in language '%s'" % (cond(B('#') in B(justSynthesize) or len(repr(line))<10,"that '"+repr(line)+"'","this line"),lastLanguage))
-                    if callSanityCheck and sanityCheck(line,lastLanguage,1): return
+                    if callGeneralCheck and generalCheck(line,lastLanguage,1): return
                     event = checkCanSynth("!synth:"+S(line)+"_"+S(lastLanguage))
                 else:
-                    if callSanityCheck and sanityCheck(text,lang,1): return
+                    if callGeneralCheck and generalCheck(text,lang,1): return
                     event = checkCanSynth(fname)
                     lastLanguage = lang
                 if not event: continue

thindown.py

@@ -1,3 +1,9 @@
+#!/usr/bin/env python
+#  (works on either Python 2 or Python 3)
+
+# program to "thin down" the gradint .py for low memory environments
+# by taking out some of the code that's unused on that platform
+
 # This file is part of the source code of Gradint
 # (c) Silas S. Brown.
 #    This program is free software; you can redistribute it and/or modify
@@ -8,11 +14,6 @@
 #    but WITHOUT ANY WARRANTY; without even the implied warranty of
 #    MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
 #    GNU General Public License for more details.
-#!/usr/bin/env python
-#  (works on either Python 2 or Python 3)
-
-# program to "thin down" the gradint .py for low memory environments
-# by taking out some of the code that's unused on that platform
 
 import sys, re
 
@@ -230,7 +231,67 @@
   to_omit = desktop_only + S60_only + android_only + android_or_S60 + not_winCE + riscos_only + mac_only
 elif "core" in sys.argv: # experimental "core code only" for 'minimal embedded porting' starting point (no UI, no synth, limited file I/O; you'll probably have to load up the event data yourself)
   version = "core"
-  to_omit = tk_only + not_S60_or_android + not_android + riscos_only + mac_only + desktop_only + winCE_only + S60_only + android_only + android_or_S60 + ["def main():","def rest_of_main():",'if __name__=="__main__":',"def transliterates_differently(text,lang):","def primitive_synthloop():","def appendVocabFileInRightLanguages():",'def delOrReplace(L2toDel,L1toDel,newL2,newL1,action="delete"):',"def sanityCheck(text,language,pauseOnError=0):","def localise(s):","def singular(number,s):","def readText(l):","def asUnicode(x):","def updateSettingsFile(fname,newVals):","def clearScreen():","def startBrowser(url):",'def getYN(msg,defaultIfEof="n"):',"def waitOnMessage(msg):","def interrupt_instructions():","def parseSynthVocab(fname,forGUI=0):","def scanSamples_inner(directory,retVal,doLimit):","def getLsDic(directory):","def check_has_variants(directory,ls):","def exec_in_a_func(x):","def scanSamples(directory=None):","def synth_from_partials(text,lang,voice=None,isStart=1):","def partials_langname(lang):","if partialsDirectory and isDirectory(partialsDirectory):",'for zipToCheck in ["yali-voice","yali-lower","cameron-voice"]:','def stripPuncEtc(text):','def can_be_synthesized(fname,dirBase=None,lang=None):','def synthcache_lookup(fname,dirBase=None,printErrors=0,justQueryCache=0,lang=None):','def textof(fname):','if synthCache and transTbl in synthCache_contents:','if synthCache:','class Partials_Synth(Synth):','def abspath_from_start(p):','class SynthEvent(Event):','def pinyin_uColon_to_V(pinyin):','def synth_event(language,text,is_prompt=0):','def get_synth_if_possible(language,warn=1,to_transliterate=False):','if wavPlayer_override or (unix and not macsound and not (oss_sound_device=="/dev/sound/dsp" or oss_sound_device=="/dev/dsp")):','def fix_compatibility(utext):','def read_chinese_number(num):','def preprocess_chinese_numbers(utext,isCant=0):','def intor0(v):','def fix_pinyin(pinyin,en_words):','def fix_commas(text):','def shell_escape(text):','class SimpleZhTransliterator(object):','def sort_out_pinyin_3rd_tones(pinyin):','def ensure_unicode(text):','def unzip_and_delete(f,specificFiles="",ignore_fail=0):','class Synth(object):','def quickGuess(letters,lettersPerSec):',"def changeToDirOf(file,winsound_also=0):",'if app or appuifw or android:','def subst_some_synth_for_synthcache(events):','def decide_subst_synth(cache_fname):','if winsound or winCEsound or mingw32 or riscos_sound or not hasattr(os,"tempnam") or android:','if len(sys.argv)>1:','def readSettings(f):','def exc_info(inGradint=True):','if not fileExists(configFiles[0]):','def u8strip(d):',]
+  to_omit = tk_only + not_S60_or_android + not_android + riscos_only + mac_only + desktop_only + winCE_only + S60_only + android_only + android_or_S60 + [
+"def main():",
+"def rest_of_main():",
+'if __name__=="__main__":',
+"def transliterates_differently(text,lang):",
+"def primitive_synthloop():",
+"def appendVocabFileInRightLanguages():",
+'def delOrReplace(L2toDel,L1toDel,newL2,newL1,action="delete"):',
+"def generalCheck(text,language,pauseOnError=0):",
+"def localise(s):",
+"def singular(number,s):",
+"def readText(l):",
+"def asUnicode(x):",
+"def updateSettingsFile(fname,newVals):",
+"def clearScreen():",
+"def startBrowser(url):",'def getYN(msg,defaultIfEof="n"):',"def waitOnMessage(msg):",
+"def interrupt_instructions():",
+"def parseSynthVocab(fname,forGUI=0):",
+"def scanSamples_inner(directory,retVal,doLimit):",
+"def getLsDic(directory):",
+"def check_has_variants(directory,ls):",
+"def exec_in_a_func(x):",
+"def scanSamples(directory=None):",
+"def synth_from_partials(text,lang,voice=None,isStart=1):",
+"def partials_langname(lang):",
+"if partialsDirectory and isDirectory(partialsDirectory):",
+'for zipToCheck in ["yali-voice","yali-lower","cameron-voice"]:',
+'def stripPuncEtc(text):',
+'def can_be_synthesized(fname,dirBase=None,lang=None):',
+'def synthcache_lookup(fname,dirBase=None,printErrors=0,justQueryCache=0,lang=None):',
+'def textof(fname):',
+'if synthCache and transTbl in synthCache_contents:',
+'if synthCache:',
+'class Partials_Synth(Synth):',
+'def abspath_from_start(p):',
+'class SynthEvent(Event):',
+'def pinyin_uColon_to_V(pinyin):',
+'def synth_event(language,text,is_prompt=0):',
+'def get_synth_if_possible(language,warn=1,to_transliterate=False):',
+'if wavPlayer_override or (unix and not macsound and not (oss_sound_device=="/dev/sound/dsp" or oss_sound_device=="/dev/dsp")):',
+'def fix_compatibility(utext):',
+'def read_chinese_number(num):',
+'def preprocess_chinese_numbers(utext,isCant=0):',
+'def intor0(v):',
+'def fix_pinyin(pinyin,en_words):',
+'def fix_commas(text):',
+'def shell_escape(text):',
+'class SimpleZhTransliterator(object):',
+'def sort_out_pinyin_3rd_tones(pinyin):',
+'def ensure_unicode(text):',
+'def unzip_and_delete(f,specificFiles="",ignore_fail=0):',
+'class Synth(object):',
+'def quickGuess(letters,lettersPerSec):',"def changeToDirOf(file,winsound_also=0):",'if app or appuifw or android:',
+'def subst_some_synth_for_synthcache(events):',
+'def decide_subst_synth(cache_fname):',
+'if winsound or winCEsound or mingw32 or riscos_sound or not hasattr(os,"tempnam") or android:',
+'if len(sys.argv)>1:',
+'def readSettings(f):',
+'def exc_info(inGradint=True):',
+'if not fileExists(configFiles[0]):',
+'def u8strip(d):']
 else: assert 0, "Unrecognised version on command line"
 
 revertToIndent = lastIndentLevel = indentLevel = -1