This is Richie Hindle's patch

[ 643835 ] Set Next Statement for Python debuggers

with a few tweaks by me: adding an unsigned or two, mentioning that
not all jumps are allowed in the doc for pdb, adding a NEWS item and
a note to whatsnew, and AuCTeX doing something cosmetic to libpdb.tex.

1 files changed, 259 insertions, 1 deletions

diff --git a/Objects/frameobject.c b/Objects/frameobject.c
index 3036ab6..b7b3021 100644
--- a/Objects/frameobject.c
+++ b/Objects/frameobject.c
@@ -8,6 +8,9 @@
 #include "opcode.h"
 #include "structmember.h"
 
+#define MIN(a, b) ((a) < (b) ? (a) : (b))
+#define MAX(a, b) ((a) > (b) ? (a) : (b))
+
 #define OFF(x) offsetof(PyFrameObject, x)
 
 static PyMemberDef frame_memberlist[] = {
@@ -44,6 +47,260 @@ frame_getlineno(PyFrameObject *f, void *closure)
 	return PyInt_FromLong(lineno);
 }
 
+/* Setter for f_lineno - you can set f_lineno from within a trace function in
+ * order to jump to a given line of code, subject to some restrictions.  Most
+ * lines are OK to jump to because they don't make any assumptions about the
+ * state of the stack (obvious because you could remove the line and the code
+ * would still work without any stack errors), but there are some constructs
+ * that limit jumping:
+ *
+ *  o Lines with an 'except' statement on them can't be jumped to, because
+ *    they expect an exception to be on the top of the stack.
+ *  o Lines that live in a 'finally' block can't be jumped from or to, since
+ *    the END_FINALLY expects to clean up the stack after the 'try' block.
+ *  o 'try'/'for'/'while' blocks can't be jumped into because the blockstack
+ *    needs to be set up before their code runs, and for 'for' loops the
+ *    iterator needs to be on the stack.
+ */
+static int
+frame_setlineno(PyFrameObject *f, PyObject* p_new_lineno)
+{
+	int new_lineno = 0;		/* The new value of f_lineno */
+	int new_lasti = 0;		/* The new value of f_lasti */
+	int new_iblock = 0;		/* The new value of f_iblock */
+	char *code = NULL;		/* The bytecode for the frame... */
+	int code_len = 0;		/* ...and its length */
+	char *lnotab = NULL;		/* Iterating over co_lnotab */
+	int lnotab_len = 0;		/* (ditto) */
+	int offset = 0;			/* (ditto) */
+	int line = 0;			/* (ditto) */
+	int addr = 0;			/* (ditto) */
+	int min_addr = 0;		/* Scanning the SETUPs and POPs */
+	int max_addr = 0;		/* (ditto) */
+	int delta_iblock = 0;		/* (ditto) */
+	int min_delta_iblock = 0;	/* (ditto) */
+	int min_iblock = 0;		/* (ditto) */
+	int f_lasti_setup_addr = 0;	/* Policing no-jump-into-finally */
+	int new_lasti_setup_addr = 0;	/* (ditto) */
+	int blockstack[CO_MAXBLOCKS];	/* Walking the 'finally' blocks */
+	int in_finally[CO_MAXBLOCKS];	/* (ditto) */
+	int blockstack_top = 0;		/* (ditto) */
+	int setup_op = 0;               /* (ditto) */
+
+	/* f_lineno must be an integer. */
+	if (!PyInt_Check(p_new_lineno)) {
+		PyErr_SetString(PyExc_ValueError,
+				"lineno must be an integer");
+		return -1;
+	}
+
+	/* You can only do this from within a trace function, not via
+	 * _getframe or similar hackery. */
+	if (!f->f_trace)
+	{
+		PyErr_Format(PyExc_ValueError,
+			     "f_lineno can only be set by a trace function");
+		return -1;
+	}
+
+	/* Fail if the line comes before the start of the code block. */
+	new_lineno = (int) PyInt_AsLong(p_new_lineno);
+	if (new_lineno < f->f_code->co_firstlineno) {
+		PyErr_Format(PyExc_ValueError,
+			     "line %d comes before the current code block",
+			     new_lineno);
+		return -1;
+	}
+
+	/* Find the bytecode offset for the start of the given line, or the
+	 * first code-owning line after it. */
+	PyString_AsStringAndSize(f->f_code->co_lnotab, &lnotab, &lnotab_len);
+	addr = 0;
+	line = f->f_code->co_firstlineno;
+	new_lasti = -1;
+	for (offset = 0; offset < lnotab_len; offset += 2) {
+		addr += lnotab[offset];
+		line += lnotab[offset+1];
+		if (line >= new_lineno) {
+			new_lasti = addr;
+			new_lineno = line;
+			break;
+		}
+	}
+
+	/* If we didn't reach the requested line, return an error. */
+	if (new_lasti == -1) {
+		PyErr_Format(PyExc_ValueError,
+			     "line %d comes after the current code block",
+			     new_lineno);
+		return -1;
+	}
+
+	/* We're now ready to look at the bytecode. */
+	PyString_AsStringAndSize(f->f_code->co_code, &code, &code_len);
+	min_addr = MIN(new_lasti, f->f_lasti);
+	max_addr = MAX(new_lasti, f->f_lasti);
+
+	/* You can't jump onto a line with an 'except' statement on it -
+	 * they expect to have an exception on the top of the stack, which
+	 * won't be true if you jump to them.  They always start with code
+	 * that either pops the exception using POP_TOP (plain 'except:'
+	 * lines do this) or duplicates the exception on the stack using
+	 * DUP_TOP (if there's an exception type specified).  See compile.c,
+	 * 'com_try_except' for the full details.  There aren't any other
+	 * cases (AFAIK) where a line's code can start with DUP_TOP or
+	 * POP_TOP, but if any ever appear, they'll be subject to the same
+	 * restriction (but with a different error message). */
+	if (code[new_lasti] == DUP_TOP || code[new_lasti] == POP_TOP) {
+		PyErr_SetString(PyExc_ValueError,
+		    "can't jump to 'except' line as there's no exception");
+		return -1;
+	}
+
+	/* You can't jump into or out of a 'finally' block because the 'try'
+	 * block leaves something on the stack for the END_FINALLY to clean
+	 * up.  So we walk the bytecode, maintaining a simulated blockstack.
+	 * When we reach the old or new address and it's in a 'finally' block
+	 * we note the address of the corresponding SETUP_FINALLY.  The jump
+	 * is only legal if neither address is in a 'finally' block or
+	 * they're both in the same one.  'blockstack' is a stack of the
+	 * bytecode addresses of the SETUP_X opcodes, and 'in_finally' tracks
+	 * whether we're in a 'finally' block at each blockstack level. */
+	f_lasti_setup_addr = -1;
+	new_lasti_setup_addr = -1;
+	memset(blockstack, '\0', sizeof(blockstack));
+	memset(in_finally, '\0', sizeof(in_finally));
+	blockstack_top = 0;
+	for (addr = 0; addr < code_len; addr++) {
+		unsigned char op = code[addr];
+		switch (op) {
+		case SETUP_LOOP:
+		case SETUP_EXCEPT:
+		case SETUP_FINALLY:
+			blockstack[blockstack_top++] = addr;
+			in_finally[blockstack_top-1] = 0;
+			break;
+
+		case POP_BLOCK:
+			setup_op = code[blockstack[blockstack_top-1]];
+			if (setup_op == SETUP_FINALLY) {
+				in_finally[blockstack_top-1] = 1;
+			}
+			else {
+				blockstack_top--;
+			}
+			break;
+
+		case END_FINALLY:
+			/* Ignore END_FINALLYs for SETUP_EXCEPTs - they exist
+			 * in the bytecode but don't correspond to an actual
+			 * 'finally' block. */
+			setup_op = code[blockstack[blockstack_top-1]];
+			if (setup_op == SETUP_FINALLY) {
+				blockstack_top--;
+			}
+			break;
+		}
+
+		/* For the addresses we're interested in, see whether they're
+		 * within a 'finally' block and if so, remember the address
+		 * of the SETUP_FINALLY. */
+		if (addr == new_lasti || addr == f->f_lasti) {
+			int i = 0;
+			int setup_addr = -1;
+			for (i = blockstack_top-1; i >= 0; i--) {
+				if (in_finally[i]) {
+					setup_addr = blockstack[i];
+					break;
+				}
+			}
+
+			if (setup_addr != -1) {
+				if (addr == new_lasti) {
+					new_lasti_setup_addr = setup_addr;
+				}
+
+				if (addr == f->f_lasti) {
+					f_lasti_setup_addr = setup_addr;
+				}
+			}
+		}
+
+		if (op >= HAVE_ARGUMENT) {
+			addr += 2;
+		}
+	}
+
+	if (new_lasti_setup_addr != f_lasti_setup_addr) {
+		PyErr_SetString(PyExc_ValueError,
+			    "can't jump into or out of a 'finally' block");
+		return -1;
+	}
+
+
+	/* Police block-jumping (you can't jump into the middle of a block)
+	 * and ensure that the blockstack finishes up in a sensible state (by
+	 * popping any blocks we're jumping out of).  We look at all the
+	 * blockstack operations between the current position and the new
+	 * one, and keep track of how many blocks we drop out of on the way.
+	 * By also keeping track of the lowest blockstack position we see, we
+	 * can tell whether the jump goes into any blocks without coming out
+	 * again - in that case we raise an exception below. */
+	delta_iblock = 0;
+	for (addr = min_addr; addr < max_addr; addr++) {
+		unsigned char op = code[addr];
+		switch (op) {
+		case SETUP_LOOP:
+		case SETUP_EXCEPT:
+		case SETUP_FINALLY:
+			delta_iblock++;
+			break;
+
+		case POP_BLOCK:
+			delta_iblock--;
+			break;
+		}
+
+		min_delta_iblock = MIN(min_delta_iblock, delta_iblock);
+
+		if (op >= HAVE_ARGUMENT) {
+			addr += 2;
+		}
+	}
+
+	/* Derive the absolute iblock values from the deltas. */
+	min_iblock = f->f_iblock + min_delta_iblock;
+	if (new_lasti > f->f_lasti) {
+		/* Forwards jump. */
+		new_iblock = f->f_iblock + delta_iblock;
+	}
+	else {
+		/* Backwards jump. */
+		new_iblock = f->f_iblock - delta_iblock;
+	}
+
+	/* Are we jumping into a block? */
+	if (new_iblock > min_iblock) {
+		PyErr_SetString(PyExc_ValueError,
+				"can't jump into the middle of a block");
+		return -1;
+	}
+
+	/* Pop any blocks that we're jumping out of. */
+	while (f->f_iblock > new_iblock) {
+		PyTryBlock *b = &f->f_blockstack[--f->f_iblock];
+		while ((f->f_stacktop - f->f_valuestack) > b->b_level) {
+			PyObject *v = (*--f->f_stacktop);
+			Py_DECREF(v);
+		}
+	}
+
+	/* Finally set the new f_lineno and f_lasti and return OK. */
+	f->f_lineno = new_lineno;
+	f->f_lasti = new_lasti;
+	return 0;
+}
+
 static PyObject *
 frame_gettrace(PyFrameObject *f, void *closure)
 {
@@ -77,7 +334,8 @@ frame_settrace(PyFrameObject *f, PyObject* v, void *closure)
 
 static PyGetSetDef frame_getsetlist[] = {
 	{"f_locals",	(getter)frame_getlocals, NULL, NULL},
-	{"f_lineno",	(getter)frame_getlineno, NULL, NULL},
+	{"f_lineno",	(getter)frame_getlineno,
+			(setter)frame_setlineno, NULL},
 	{"f_trace",	(getter)frame_gettrace, (setter)frame_settrace, NULL},
 	{0}
 };