Menu
▾
▴
[Jython-checkins] CVS: jython/org/python/modules/sre PatternObject.java,1.7,1.8 SRE_STATE.java,1.7,1.8 ScannerObject.java,1.3,1.4
|
From: Finn B. <bc...@us...> - 2002-01-06 15:53:27
|
Update of /cvsroot/jython/jython/org/python/modules/sre
In directory usw-pr-cvs1:/tmp/cvs-serv21026/sre
Modified Files:
PatternObject.java SRE_STATE.java ScannerObject.java
Log Message:
CPython-2.2's sre changes to MAGIC==20010701.
Index: PatternObject.java
===================================================================
RCS file: /cvsroot/jython/jython/org/python/modules/sre/PatternObject.java,v
retrieving revision 1.7
retrieving revision 1.8
diff -C2 -d -r1.7 -r1.8
*** PatternObject.java 2001/11/27 19:07:22 1.7
--- PatternObject.java 2002/01/06 15:53:25 1.8
***************
*** 27,30 ****
--- 27,31 ----
public int flags;
org.python.core.PyObject indexgroup;
+ public int codesize;
***************
*** 32,38 ****
int groups, PyObject groupindex, PyObject indexgroup) {
! this.pattern = pattern.toString();
this.flags = flags;
this.code = code;
this.groups = groups;
this.groupindex = groupindex;
--- 33,41 ----
int groups, PyObject groupindex, PyObject indexgroup) {
! if (pattern != null)
! this.pattern = pattern.toString();
this.flags = flags;
this.code = code;
+ this.codesize = code.length;
this.groups = groups;
this.groupindex = groupindex;
***************
*** 82,93 ****
int count = ap.getInt(2, 0);
! return call("_sub", new PyObject[] {
! Py.java2py(this),
! template,
! Py.newString(string),
! Py.newInteger(count) });
}
public PyObject subn(PyObject[] args, String[] kws) {
ArgParser ap = new ArgParser("subn", args, kws,
--- 85,93 ----
int count = ap.getInt(2, 0);
! return subx(template, string, count, false);
}
+
public PyObject subn(PyObject[] args, String[] kws) {
ArgParser ap = new ArgParser("subn", args, kws,
***************
*** 97,105 ****
int count = ap.getInt(2, 0);
! return call("_subn", new PyObject[] {
! Py.java2py(this),
! template,
! Py.newString(string),
! Py.newInteger(count) });
}
--- 97,183 ----
int count = ap.getInt(2, 0);
! return subx(template, string, count, true);
! }
!
!
! private PyObject subx(PyObject template, String string, int count,
! boolean subn)
! {
! PyObject filter = null;
! boolean filter_is_callable = false;
! if (template.isCallable()) {
! filter = template;
! filter_is_callable = true;
! } else {
! boolean literal = false;
! if (template instanceof PyString) {
! literal = template.toString().indexOf('\\') < 0;
! }
! if (literal) {
! filter = template;
! filter_is_callable = false;
! } else {
! filter = call("sre", "_subx", new PyObject[] {
! this, template});
! filter_is_callable = filter.isCallable();
! }
! }
!
! SRE_STATE state = new SRE_STATE(string, 0, Integer.MAX_VALUE, flags);
!
! StringBuffer buf = new StringBuffer();
!
! int n = 0;
! int i = 0;
!
! while (count == 0 || n < count) {
! state.state_reset();
! state.ptr = state.start;
! int status = state.SRE_SEARCH(code, 0);
! if (status <= 0) {
! if (status == 0)
! break;
! _error(status);
! }
! int b = state.start;
! int e = state.ptr;
!
! if (i < b) {
! /* get segment before this match */
! buf.append(string.substring(i, b));
! }
! if (! (i == b && i == e && n > 0)) {
! PyObject item;
! if (filter_is_callable) {
! /* pass match object through filter */
! MatchObject match = _pattern_new_match(state, string, 1);
! item = filter.__call__(match);
! } else {
! item = filter;
! }
!
! if (item != Py.None) {
! buf.append(item.toString());
! }
! i = e;
! n++;
! }
!
! /* move on */
! if (state.ptr == state.start)
! state.start = state.ptr + 1;
! else
! state.start = state.ptr;
! }
! if (i < state.endpos) {
! buf.append(string.substring(i, state.endpos));
! }
!
! if (subn)
! return new PyTuple(new PyObject[] {
! Py.newString(buf.toString()), Py.newInteger(n)
! });
! else
! return Py.newString(buf.toString());
}
***************
*** 109,122 ****
"source", "maxsplit");
String string = ap.getString(0);
! int count = ap.getInt(1, 0);
! return call("_split", new PyObject[] {
! Py.java2py(this),
! Py.newString(string),
! Py.newInteger(count) });
}
! private PyObject call(String function, PyObject[] args) {
! PyObject sre = imp.importName("sre", true);
return sre.invoke(function, args);
}
--- 187,238 ----
"source", "maxsplit");
String string = ap.getString(0);
! int maxsplit = ap.getInt(1, 0);
! SRE_STATE state = new SRE_STATE(string, 0, Integer.MAX_VALUE, flags);
!
! PyList list = new PyList();
!
! int n = 0;
! int last = state.start;
! while (maxsplit == 0 || n < maxsplit) {
! state.state_reset();
! state.ptr = state.start;
! int status = state.SRE_SEARCH(code, 0);
! if (status <= 0) {
! if (status == 0)
! break;
! _error(status);
! }
! if (state.start == state.ptr) {
! if (last == state.end)
! break;
! /* skip one character */
! state.start = state.ptr + 1;
! continue;
! }
!
! /* get segment before this match */
! PyObject item = Py.newString(string.substring(last, state.start));
! list.append(item);
!
! for (int i = 0; i < groups; i++) {
! String s = state.getslice(i+1, string, false);
! if (s != null)
! list.append(Py.newString(s));
! else
! list.append(Py.None);
! }
! n += 1;
! last = state.start = state.ptr;
! }
!
! PyObject item = Py.newString(string.substring(last, state.endpos));
! list.append(item);
!
! return list;
}
! private PyObject call(String module, String function, PyObject[] args) {
! PyObject sre = imp.importName(module, true);
return sre.invoke(function, args);
}
***************
*** 149,158 ****
break;
case 1:
! item = Py.newString(state.getslice(1, string));
break;
default:
PyObject[] t = new PyObject[groups];
for (int i = 0; i < groups; i++)
! t[i] = Py.newString(state.getslice(i+1, string));
item = new PyTuple(t);
break;
--- 265,274 ----
break;
case 1:
! item = Py.newString(state.getslice(1, string, true));
break;
default:
PyObject[] t = new PyObject[groups];
for (int i = 0; i < groups; i++)
! t[i] = Py.newString(state.getslice(i+1, string, true));
item = new PyTuple(t);
break;
***************
*** 176,179 ****
--- 292,311 ----
}
+
+ /* Enable "finditer" when iter support is added.
+ public PyObject finditer(String string) {
+ return finditer(string, 0, Integer.MAX_VALUE);
+ }
+
+ public PyObject finditer(String string, int start) {
+ return finditer(string, start, Integer.MAX_VALUE);
+ }
+
+ public PyObject finditer(String string, int start, int end) {
+ ScannerObject scanner = scanner(string, start, end);
+ PyObject search = scanner.__findattr__("search");
+ return new PyCallIterator(search, Py.None);
+ }
+ */
public ScannerObject scanner(String string) {
Index: SRE_STATE.java
===================================================================
RCS file: /cvsroot/jython/jython/org/python/modules/sre/SRE_STATE.java,v
retrieving revision 1.7
retrieving revision 1.8
diff -C2 -d -r1.7 -r1.8
*** SRE_STATE.java 2001/11/27 19:07:22 1.7
--- SRE_STATE.java 2002/01/06 15:53:25 1.8
***************
*** 41,62 ****
public static final int SRE_OP_CATEGORY = 9;
public static final int SRE_OP_CHARSET = 10;
! public static final int SRE_OP_GROUPREF = 11;
! public static final int SRE_OP_GROUPREF_IGNORE = 12;
! public static final int SRE_OP_IN = 13;
! public static final int SRE_OP_IN_IGNORE = 14;
! public static final int SRE_OP_INFO = 15;
! public static final int SRE_OP_JUMP = 16;
! public static final int SRE_OP_LITERAL = 17;
! public static final int SRE_OP_LITERAL_IGNORE = 18;
! public static final int SRE_OP_MARK = 19;
! public static final int SRE_OP_MAX_UNTIL = 20;
! public static final int SRE_OP_MIN_UNTIL = 21;
! public static final int SRE_OP_NOT_LITERAL = 22;
! public static final int SRE_OP_NOT_LITERAL_IGNORE = 23;
! public static final int SRE_OP_NEGATE = 24;
! public static final int SRE_OP_RANGE = 25;
! public static final int SRE_OP_REPEAT = 26;
! public static final int SRE_OP_REPEAT_ONE = 27;
! public static final int SRE_OP_SUBPATTERN = 28;
public static final int SRE_AT_BEGINNING = 0;
--- 41,63 ----
public static final int SRE_OP_CATEGORY = 9;
public static final int SRE_OP_CHARSET = 10;
! public static final int SRE_OP_BIGCHARSET = 11;
! public static final int SRE_OP_GROUPREF = 12;
! public static final int SRE_OP_GROUPREF_IGNORE = 13;
! public static final int SRE_OP_IN = 14;
! public static final int SRE_OP_IN_IGNORE = 15;
! public static final int SRE_OP_INFO = 16;
! public static final int SRE_OP_JUMP = 17;
! public static final int SRE_OP_LITERAL = 18;
! public static final int SRE_OP_LITERAL_IGNORE = 19;
! public static final int SRE_OP_MARK = 20;
! public static final int SRE_OP_MAX_UNTIL = 21;
! public static final int SRE_OP_MIN_UNTIL = 22;
! public static final int SRE_OP_NOT_LITERAL = 23;
! public static final int SRE_OP_NOT_LITERAL_IGNORE = 24;
! public static final int SRE_OP_NEGATE = 25;
! public static final int SRE_OP_RANGE = 26;
! public static final int SRE_OP_REPEAT = 27;
! public static final int SRE_OP_REPEAT_ONE = 28;
! public static final int SRE_OP_SUBPATTERN = 29;
public static final int SRE_AT_BEGINNING = 0;
***************
*** 329,332 ****
--- 330,344 ----
break;
+ case SRE_OP_BIGCHARSET:
+ /* <BIGCHARSET> <blockcount> <256 blockindices> <blocks> */
+ int count = set[setidx++];
+ int block = set[ch >> 8];
+ setidx += 128;
+ int idx = block*16 + ((ch & 255)>>4);
+ if ((set[setidx + idx] & (1 << (ch & 15))) != 0)
+ return ok;
+ setidx += count*16;
+ break;
+
case SRE_OP_CATEGORY:
/* <CATEGORY> <code> */
***************
*** 816,819 ****
--- 828,832 ----
return i;
mark_restore(0, lastmark);
+ this.lastmark = lastmark;
rp.count = count - 1;
this.ptr = ptr;
***************
*** 859,872 ****
/* see if the tail matches */
this.repeat = rp.prev;
! if (pattern[rp.pidx + 2] == 65535) {
! /* unbounded repeat */
! for (;;) {
! i = SRE_MATCH(pattern, pidx, level + 1);
! if (i != 0 || ptr >= end)
! break;
! this.ptr = ++ptr;
! }
! } else
! i = SRE_MATCH(pattern, pidx, level + 1);
if (i != 0)
return i;
--- 872,876 ----
/* see if the tail matches */
this.repeat = rp.prev;
! i = SRE_MATCH(pattern, pidx, level + 1);
if (i != 0)
return i;
***************
*** 991,994 ****
--- 995,1000 ----
this.start = ptr;
this.ptr = ++ptr;
+ if ((flags & SRE_INFO_LITERAL) != 0)
+ return 1;
status = SRE_MATCH(pattern, pidx + 2, 1);
if (status != 0)
***************
*** 1028,1035 ****
-
-
-
-
final boolean sre_category(char category, char ch) {
switch (category) {
--- 1034,1037 ----
***************
*** 1175,1179 ****
! String getslice(int index, String string) {
int i, j;
--- 1177,1181 ----
! String getslice(int index, String string, boolean empty) {
int i, j;
***************
*** 1181,1185 ****
if (string == null || mark[index] == -1 || mark[index+1] == -1) {
! i = j = 0;
} else {
i = mark[index];
--- 1183,1192 ----
if (string == null || mark[index] == -1 || mark[index+1] == -1) {
! if (empty) {
! /* want empty string */
! i = j = 0;
! } else {
! return null;
! }
} else {
i = mark[index];
Index: ScannerObject.java
===================================================================
RCS file: /cvsroot/jython/jython/org/python/modules/sre/ScannerObject.java,v
retrieving revision 1.3
retrieving revision 1.4
diff -C2 -d -r1.3 -r1.4
*** ScannerObject.java 2001/11/27 13:51:37 1.3
--- ScannerObject.java 2002/01/06 15:53:25 1.4
***************
*** 16,20 ****
package org.python.modules.sre;
! public class ScannerObject {
PatternObject pattern;
String string;
--- 16,22 ----
package org.python.modules.sre;
! import org.python.core.*;
!
! public class ScannerObject extends PyObject {
PatternObject pattern;
String string;
|
