Guido van Rossum | 5dd52d3 | 1995-04-10 11:47:11 +0000 | [diff] [blame] | 1 | # Utility module for 'icache.py': interpret tag tables and indirect nodes. |
| 2 | |
| 3 | # (This module is a bit chatty when confronted with the unexpected.) |
| 4 | |
| 5 | |
| 6 | import regexp |
| 7 | import string |
| 8 | import ifile |
| 9 | |
| 10 | |
| 11 | # Get the tag table of an open file, as a dictionary. |
| 12 | # Seeks around in the file; after reading, the position is undefined. |
| 13 | # Return an empty tag table if none is found. |
| 14 | # |
| 15 | def get_tags(f): |
| 16 | # |
| 17 | # First see if the last "node" is the end of tag table marker. |
| 18 | # |
| 19 | f.seek(0, 2) # Seek to EOF |
| 20 | end = f.tell() |
| 21 | buf = ifile.backup_node(f, end) |
| 22 | if not labelmatch(buf, 0, 'end tag table\n'): |
| 23 | return {} # No succes |
| 24 | # |
| 25 | # Next backup to the previous "node" -- the tag table itself. |
| 26 | # |
| 27 | ###print 'Getting prebuilt tag table...' |
| 28 | end = f.tell() - len(buf) |
| 29 | buf = ifile.backup_node(f, end) |
| 30 | label = 'tag table:\n' |
| 31 | if not labelmatch(buf, 0, label): |
| 32 | print 'Weird: end tag table marker but no tag table?' |
| 33 | print 'Node begins:', `buf[:50]` |
| 34 | return {} |
| 35 | # |
| 36 | # Now read the whole tag table. |
| 37 | # |
| 38 | end = f.tell() - len(buf) # Do this first! |
| 39 | buf = ifile.read_node(f, buf) |
| 40 | # |
| 41 | # First check for an indirection table. |
| 42 | # |
| 43 | indirlist = [] |
| 44 | if labelmatch(buf, len(label), '(indirect)\n'): |
| 45 | indirbuf = ifile.backup_node(f, end) |
| 46 | if not labelmatch(indirbuf, 0, 'indirect:\n'): |
| 47 | print 'Weird: promised indirection table not found' |
| 48 | print 'Node begins:', `indirbuf[:50]` |
| 49 | # Carry on. Things probably won't work though. |
| 50 | else: |
| 51 | indirbuf = ifile.read_node(f, indirbuf) |
| 52 | indirlist = parse_indirlist(indirbuf) |
| 53 | # |
| 54 | # Now parse the tag table. |
| 55 | # |
| 56 | findtag = regexp.compile('^(.*[nN]ode:[ \t]*(.*))\177([0-9]+)$').match |
| 57 | i = 0 |
| 58 | tags = {} |
| 59 | while 1: |
| 60 | match = findtag(buf, i) |
| 61 | if not match: |
| 62 | break |
| 63 | (a,b), (a1,b1), (a2,b2), (a3,b3) = match |
| 64 | i = b |
| 65 | line = buf[a1:b1] |
| 66 | node = string.lower(buf[a2:b2]) |
| 67 | offset = eval(buf[a3:b3]) # XXX What if it overflows? |
| 68 | if tags.has_key(node): |
| 69 | print 'Duplicate key in tag table:', `node` |
| 70 | file, offset = map_offset(offset, indirlist) |
| 71 | tags[node] = file, offset, line |
| 72 | # |
| 73 | return tags |
| 74 | |
| 75 | |
| 76 | # Return true if buf[i:] begins with a label, after lower case conversion. |
| 77 | # The label argument must be in lower case. |
| 78 | # |
| 79 | def labelmatch(buf, i, label): |
| 80 | return string.lower(buf[i:i+len(label)]) == label |
| 81 | |
| 82 | |
| 83 | # Parse the indirection list. |
| 84 | # Return a list of (filename, offset) pairs ready for use. |
| 85 | # |
| 86 | def parse_indirlist(buf): |
| 87 | list = [] |
| 88 | findindir = regexp.compile('^(.+):[ \t]*([0-9]+)$').match |
| 89 | i = 0 |
| 90 | while 1: |
| 91 | match = findindir(buf, i) |
| 92 | if not match: |
| 93 | break |
| 94 | (a,b), (a1,b1), (a2,b2) = match |
| 95 | file = buf[a1:b1] |
| 96 | offset = eval(buf[a2:b2]) # XXX What if this gets overflow? |
| 97 | list.append(file, offset) |
| 98 | i = b |
| 99 | return list |
| 100 | |
| 101 | |
| 102 | # Map an offset through the indirection list. |
| 103 | # Return (filename, new_offset). |
| 104 | # If the list is empty, return the given offset and an empty file name. |
| 105 | # |
| 106 | def map_offset(offset, indirlist): |
| 107 | if not indirlist: |
| 108 | return '', offset |
| 109 | # |
| 110 | # XXX This could be done more elegant. |
| 111 | # |
| 112 | filex, offx = indirlist[0] |
| 113 | for i in range(len(indirlist)): |
| 114 | file1, off1 = indirlist[i] |
| 115 | if i+1 >= len(indirlist): |
| 116 | file2, off2 = '', 0x7fffffff |
| 117 | else: |
| 118 | file2, off2 = indirlist[i+1] |
| 119 | if off1 <= offset < off2: |
| 120 | # Add offx+2 to compensate for extra header. |
| 121 | # No idea whether this is always correct. |
| 122 | return file1, offset-off1 + offx+2 |
| 123 | # |
| 124 | # XXX Shouldn't get here. |
| 125 | # |
| 126 | print 'Oops, map_offset fell through' |
| 127 | return '', offset # Not likely to get good results |