Grammalecte  Check-in [709b969c75]

Overview
Comment:[build][bug] graph builder: action identifier must be more specific to avoid rules confusion
Downloads: Tarball | ZIP archive | SQL archive
Timelines: family | ancestors | descendants | both | build | rg
Files: files | file ages | folders
SHA3-256: 709b969c75c92ea49db6a52b945b473b0744b06ccd95a50b5a0ae96c67acd5fb
User & Date: olr on 2018-08-11 14:00:17
Other Links: branch diff | manifest | tags
Context
2018-08-11
19:22
[fr] conversion: regex rules -> graph rules check-in: e620deee31 user: olr tags: fr, rg
14:00
[build][bug] graph builder: action identifier must be more specific to avoid rules confusion check-in: 709b969c75 user: olr tags: build, rg
10:50
[build] graph builder: function creator check-in: a8d90dbabb user: olr tags: build, rg
Changes

Modified compile_rules_graph.py from [f91ed363c6] to [cb7c6efd58].

14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
...
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145



146
147
148
149
150
151
152
dFUNCTIONS = {}
dFUNCNAME = {}


def createFunction (sType, sActionId, sCode, bStartWithEqual=False):
    "create a function (stored in dFUNCTIONS) and return function name"
    sCode = prepareFunction(sCode)
    if sActionId not in dFUNCNAME:
        dFUNCNAME[sActionId] = {}
    if sCode not in dFUNCNAME[sActionId]:
        dFUNCNAME[sActionId][sCode] = len(dFUNCNAME[sActionId])+1
    sFuncName = "_g_" + sType + "_" + sActionId + "_" + str(dFUNCNAME[sActionId][sCode])
    dFUNCTIONS[sFuncName] = sCode
    return sFuncName  if not bStartWithEqual  else "="+sFuncName


def prepareFunction (s):
    "convert simple rule syntax to a string of Python code"
    if s[0:1] == "=":
        s = s[1:]
    s = s.replace("__also__", "bCondMemo")
    s = s.replace("__else__", "not bCondMemo")
    s = s.replace("sContext", "_sAppContext")
    s = re.sub(r"(morph|morphVC|analyse|value|displayInfo)[(]\\(\d+)", 'g_\\1(lToken[nTokenOffset+\\2]', s)
    s = re.sub(r"(morph|morphVC|analyse|value|displayInfo)[(]\\-(\d+)", 'g_\\1(lToken[nLastToken-\\2+1]', s)
    s = re.sub(r"(select|exclude|define|define_from)[(][\\](\d+)", 'g_\\1(lToken[nTokenOffset+\\2]', s)
    s = re.sub(r"(select|exclude|define|define_from)[(][\\]-(\d+)", 'g_\\1(lToken[nLastToken-\\2+1]', s)
    s = re.sub(r"(tag_before|tag_after)[(][\\](\d+)", 'g_\\1(lToken[nTokenOffset+\\2], dTags', s)
    s = re.sub(r"(tag_before|tag_after)[(][\\]-(\d+)", 'g_\\1(lToken[nLastToken-\\2+1], dTags', s)
    s = re.sub(r"space_after[(][\\](\d+)", 'g_space_between_tokens(lToken[nTokenOffset+\\1], lToken[nTokenOffset+\\1+1]', s)
    s = re.sub(r"space_after[(][\\]-(\d+)", 'g_space_between_tokens(lToken[nLastToken-\\1+1], lToken[nLastToken-\\1+2]', s)
    s = re.sub(r"analyse_with_next[(][\\](\d+)", 'g_merged_analyse(lToken[nTokenOffset+\\1], lToken[nTokenOffset+\\1+1]', s)
    s = re.sub(r"analyse_with_next[(][\\]-(\d+)", 'g_merged_analyse(lToken[nLastToken-\\1+1], lToken[nLastToken-\\1+2]', s)
    s = re.sub(r"(morph|analyse|value)\(>1", 'g_\\1(lToken[nLastToken+1]', s)                       # next token
    s = re.sub(r"(morph|analyse|value)\(<1", 'g_\\1(lToken[nTokenOffset]', s)                       # previous token
    s = re.sub(r"(morph|analyse|value)\(>(\d+)", 'g_\\1(g_token(lToken, nLastToken+\\2)', s)          # next token
    s = re.sub(r"(morph|analyse|value)\(<(\d+)", 'g_\\1(g_token(lToken, nTokenOffset+1-\\2)', s)      # previous token
    s = re.sub(r"\bspell *[(]", '_oSpellChecker.isValid(', s)
    s = re.sub(r"\bbefore\(\s*", 'look(sSentence[:lToken[1+nTokenOffset]["nStart"]], ', s)          # before(s)
    s = re.sub(r"\bafter\(\s*", 'look(sSentence[lToken[nLastToken]["nEnd"]:], ', s)                 # after(s)
    s = re.sub(r"\bbefore0\(\s*", 'look(sSentence0[:lToken[1+nTokenOffset]["nStart"]], ', s)        # before0(s)
    s = re.sub(r"\bafter0\(\s*", 'look(sSentence[lToken[nLastToken]["nEnd"]:], ', s)                # after0(s)
    s = re.sub(r"[\\](\d+)", 'lToken[nTokenOffset+\\1]["sValue"]', s)
    s = re.sub(r"[\\]-(\d+)", 'lToken[nLastToken-\\1+1]["sValue"]', s)
    return s


def genTokenLines (sTokenLine, dDef):
    "tokenize a string and return a list of lines of tokens"
    lToken = sTokenLine.split()
    lTokenLines = None
    for sToken in lToken:
................................................................................
    "generator: create rule as list"
    # print(iLine, "//", sRuleName, "//", sTokenLine, "//", sActions, "//", nPriority)
    for lToken in genTokenLines(sTokenLine, dDef):
        # Calculate positions
        dPos = {}   # key: iGroup, value: iToken
        iGroup = 0
        #if iLine == 3971: # debug
        #    print(lToken)
        for i, sToken in enumerate(lToken):
            if sToken.startswith("(") and sToken.endswith(")"):
                lToken[i] = sToken[1:-1]
                iGroup += 1
                dPos[iGroup] = i + 1    # we add 1, for we count tokens from 1 to n (not from 0)

        # Parse actions
        for iAction, sAction in enumerate(sActions.split(" <<- ")):
            sAction = sAction.strip()
            if sAction:
                sActionId = sRuleName + "__b" + str(iActionBlock) + "_a" + str(iAction) + "_" + str(len(lToken))
                aAction = createAction(sActionId, sAction, nPriority, dOptPriority, len(lToken), dPos)
                if aAction:
                    dACTIONS[sActionId] = aAction
                    lResult = list(lToken)
                    lResult.extend(["##"+str(iLine), sActionId])



                    yield lResult
                else:
                    print(" # Error on action at line:", iLine)
                    print(sTokenLine, "\n", sActions)


def changeReferenceToken (sText, dPos):







|
|
|
|
|




|

|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|







 







|










|





>
>
>







14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
...
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
dFUNCTIONS = {}
dFUNCNAME = {}


def createFunction (sType, sActionId, sCode, bStartWithEqual=False):
    "create a function (stored in dFUNCTIONS) and return function name"
    sCode = prepareFunction(sCode)
    if sType not in dFUNCNAME:
        dFUNCNAME[sType] = {}
    if sCode not in dFUNCNAME[sType]:
        dFUNCNAME[sType][sCode] = len(dFUNCNAME[sType])+1
    sFuncName = "_g_" + sType + "_" + str(dFUNCNAME[sType][sCode])
    dFUNCTIONS[sFuncName] = sCode
    return sFuncName  if not bStartWithEqual  else "="+sFuncName


def prepareFunction (sCode):
    "convert simple rule syntax to a string of Python code"
    if sCode[0:1] == "=":
        sCode = sCode[1:]
    sCode = sCode.replace("__also__", "bCondMemo")
    sCode = sCode.replace("__else__", "not bCondMemo")
    sCode = sCode.replace("sContext", "_sAppContext")
    sCode = re.sub(r"(morph|morphVC|analyse|value|displayInfo)[(]\\(\d+)", 'g_\\1(lToken[nTokenOffset+\\2]', sCode)
    sCode = re.sub(r"(morph|morphVC|analyse|value|displayInfo)[(]\\-(\d+)", 'g_\\1(lToken[nLastToken-\\2+1]', sCode)
    sCode = re.sub(r"(select|exclude|define|define_from)[(][\\](\d+)", 'g_\\1(lToken[nTokenOffset+\\2]', sCode)
    sCode = re.sub(r"(select|exclude|define|define_from)[(][\\]-(\d+)", 'g_\\1(lToken[nLastToken-\\2+1]', sCode)
    sCode = re.sub(r"(tag_before|tag_after)[(][\\](\d+)", 'g_\\1(lToken[nTokenOffset+\\2], dTags', sCode)
    sCode = re.sub(r"(tag_before|tag_after)[(][\\]-(\d+)", 'g_\\1(lToken[nLastToken-\\2+1], dTags', sCode)
    sCode = re.sub(r"space_after[(][\\](\d+)", 'g_space_between_tokens(lToken[nTokenOffset+\\1], lToken[nTokenOffset+\\1+1]', sCode)
    sCode = re.sub(r"space_after[(][\\]-(\d+)", 'g_space_between_tokens(lToken[nLastToken-\\1+1], lToken[nLastToken-\\1+2]', sCode)
    sCode = re.sub(r"analyse_with_next[(][\\](\d+)", 'g_merged_analyse(lToken[nTokenOffset+\\1], lToken[nTokenOffset+\\1+1]', sCode)
    sCode = re.sub(r"analyse_with_next[(][\\]-(\d+)", 'g_merged_analyse(lToken[nLastToken-\\1+1], lToken[nLastToken-\\1+2]', sCode)
    sCode = re.sub(r"(morph|analyse|value)\(>1", 'g_\\1(lToken[nLastToken+1]', sCode)                       # next token
    sCode = re.sub(r"(morph|analyse|value)\(<1", 'g_\\1(lToken[nTokenOffset]', sCode)                       # previous token
    sCode = re.sub(r"(morph|analyse|value)\(>(\d+)", 'g_\\1(g_token(lToken, nLastToken+\\2)', sCode)        # next token
    sCode = re.sub(r"(morph|analyse|value)\(<(\d+)", 'g_\\1(g_token(lToken, nTokenOffset+1-\\2)', sCode)    # previous token
    sCode = re.sub(r"\bspell *[(]", '_oSpellChecker.isValid(', sCode)
    sCode = re.sub(r"\bbefore\(\s*", 'look(sSentence[:lToken[1+nTokenOffset]["nStart"]], ', sCode)          # before(sCode)
    sCode = re.sub(r"\bafter\(\s*", 'look(sSentence[lToken[nLastToken]["nEnd"]:], ', sCode)                 # after(sCode)
    sCode = re.sub(r"\bbefore0\(\s*", 'look(sSentence0[:lToken[1+nTokenOffset]["nStart"]], ', sCode)        # before0(sCode)
    sCode = re.sub(r"\bafter0\(\s*", 'look(sSentence[lToken[nLastToken]["nEnd"]:], ', sCode)                # after0(sCode)
    sCode = re.sub(r"[\\](\d+)", 'lToken[nTokenOffset+\\1]["sValue"]', sCode)
    sCode = re.sub(r"[\\]-(\d+)", 'lToken[nLastToken-\\1+1]["sValue"]', sCode)
    return sCode


def genTokenLines (sTokenLine, dDef):
    "tokenize a string and return a list of lines of tokens"
    lToken = sTokenLine.split()
    lTokenLines = None
    for sToken in lToken:
................................................................................
    "generator: create rule as list"
    # print(iLine, "//", sRuleName, "//", sTokenLine, "//", sActions, "//", nPriority)
    for lToken in genTokenLines(sTokenLine, dDef):
        # Calculate positions
        dPos = {}   # key: iGroup, value: iToken
        iGroup = 0
        #if iLine == 3971: # debug
        #    print(lToken.join(" "))
        for i, sToken in enumerate(lToken):
            if sToken.startswith("(") and sToken.endswith(")"):
                lToken[i] = sToken[1:-1]
                iGroup += 1
                dPos[iGroup] = i + 1    # we add 1, for we count tokens from 1 to n (not from 0)

        # Parse actions
        for iAction, sAction in enumerate(sActions.split(" <<- ")):
            sAction = sAction.strip()
            if sAction:
                sActionId = sRuleName + "__b" + str(iActionBlock) + "_l" + str(iLine) + "_a" + str(iAction) + "_" + str(len(lToken))
                aAction = createAction(sActionId, sAction, nPriority, dOptPriority, len(lToken), dPos)
                if aAction:
                    dACTIONS[sActionId] = aAction
                    lResult = list(lToken)
                    lResult.extend(["##"+str(iLine), sActionId])
                    if iLine == 13341:
                        print("  ".join(lToken))
                        print(sActionId, aAction)
                    yield lResult
                else:
                    print(" # Error on action at line:", iLine)
                    print(sTokenLine, "\n", sActions)


def changeReferenceToken (sText, dPos):