geometry_tools.automata.gap_parse

  1import re
  2
  3WHITESPACE = " \n\t"
  4MAX_ERRLEN = 100
  5
  6class GAPInputException(Exception):
  7    pass
  8
  9def parse_record(text):
 10    record = {}
 11    identifier_name = ""
 12
 13    i = 0
 14    while i < len(text):
 15        c = text[i]
 16        if c == ')':
 17            i += 1
 18            break
 19        elif c == ',':
 20            identifier_name = ""
 21        elif c == ":" and text[i+1] == "=":
 22            value, offset = parse_contents(text[i + 2:])
 23            record[identifier_name] = value
 24            i += offset
 25        elif c not in WHITESPACE:
 26            identifier_name += c
 27        i += 1
 28    return (record, i + 1)
 29
 30
 31def literal_contents(content_str):
 32    if re.match(r"\d*\.\d+", content_str):
 33        return float(content_str)
 34    elif re.match(r"\d+", content_str):
 35        return int(content_str)
 36    else:
 37        return content_str
 38
 39def parse_contents(text):
 40    content = ""
 41    for i, c in enumerate(text):
 42        if c == '"':
 43            content, offset = parse_quote(text[i + 1:])
 44            return (content, offset + i + 1)
 45        elif c == '[':
 46            content, offset = parse_list(text[i + 1:])
 47            return (content, offset + i + 1)
 48        elif c in WHITESPACE:
 49            continue
 50        elif c == 'r':
 51            if len(text) > i + 4 and text[i:i+4] == 'rec(':
 52                content, offset = parse_record(text[i+4:])
 53                return (content, offset + i + 4)
 54            else:
 55                content += c
 56        elif c in ",)":
 57            return (literal_contents(content), i + 1)
 58        else:
 59            content += c
 60    return (literal_contents(content), len(text))
 61
 62def parse_list(text):
 63
 64    interval = re.match(r"((-?\d+)\.\.(-?\d+)\])", text)
 65
 66    if interval and int(interval.group(2)) <= int(interval.group(3)):
 67        current_list = range(int(interval.group(2)),
 68                             int(interval.group(3)) + 1)
 69        return (current_list, len(interval.group(1)))
 70
 71    current_list = []
 72    content = ""
 73    i = 0
 74    while i < len(text):
 75        c = text[i]
 76        if c == '"':
 77            content, offset = parse_quote(text[i+1:])
 78            current_list.append(content)
 79            content = ""
 80            i += offset
 81        elif c == ',':
 82            if len(content) > 0:
 83                current_list.append(literal_contents(content))
 84            content = ""
 85        elif c == '[':
 86            newlist, offset = parse_list(text[i+1:])
 87            current_list.append(newlist)
 88            i += offset
 89        elif c == ']':
 90            if len(content) > 0:
 91                current_list.append(literal_contents(content))
 92            i += 1
 93            return (current_list, i)
 94        elif c not in WHITESPACE:
 95            content += c
 96
 97        i += 1
 98
 99    raise GAPInputException('Unclosed [: [%s'%text[:MAX_ERRLEN])
100
101def parse_quote(text):
102    #no escaped quotations
103    close_quote = text.find('"')
104    if close_quote != -1:
105        return (text[:close_quote], close_quote + 1)
106    else:
107        raise GAPInputException('Unclosed ": "%s'%text[:MAX_ERRLEN])
108
109def load_record_file(filename):
110    with open(filename, 'r') as recfile:
111        record, length = parse_record(recfile.read())
112        return record