mercurial/cext/bdiff.c
author Pierre-Yves David <pierre-yves.david@octobus.net>
Tue, 09 Apr 2024 02:54:19 +0200
changeset 51586 1cef1412af3e
parent 48821 b0dd39b91e7a
permissions -rw-r--r--
phases: rework the logic of _pushdiscoveryphase to bound complexity This rework the various graph traversal in _pushdiscoveryphase to keep the complexity in check. This is done though a couple of things: - first, limiting the space we have to explore, for example, if we are not in publishing push, we don't need to consider remote draft roots that are also draft locally, as there is nothing to be moved there. - avoid unbounded descendant computation, and use the faster "rev between" computation. This provide a massive boost to performance when exchanging with repository with a massive amount of draft, like mozilla-try: ### data-env-vars.name = mozilla-try-2023-03-22-zstd-sparse-revlog # benchmark.name = hg.command.push # bin-env-vars.hg.flavor = default # bin-env-vars.hg.py-re2-module = default # benchmark.variants.explicit-rev = all-out-heads # benchmark.variants.issue6528 = disabled # benchmark.variants.protocol = ssh # benchmark.variants.reuse-external-delta-parent = default ## benchmark.variants.revs = any-1-extra-rev before: 20.346590 seconds after: 11.232059 seconds (-38.15%, -7.48 seconds) ## benchmark.variants.revs = any-100-extra-rev before: 24.752051 seconds after: 15.367412 seconds (-37.91%, -9.38 seconds) After this changes, the push operation is still quite too slow. Some of this can be attributed to general phases slowness (reading all the roots from disk for example) and other know slowness (not using persistent-nodemap, branchmap, tags, etc. We are also working on them, but with this series, phase discovery during push no longer showing up in profile and this is a pretty nice and bit low-hanging fruit out of the way. ### (same case as the above) # benchmark.variants.revs = any-1-extra-rev pre-%ln-change: 44.235070 this-changeset: 11.232059 seconds (-74.61%, -33.00 seconds) # benchmark.variants.revs = any-100-extra-rev pre-%ln-change: 49.234697 this-changeset: 15.367412 seconds (-68.79%, -33.87 seconds) Note that with this change, the `hg push` performance is now much closer to the `hg pull` performance, even it still lagging behind a bit. (and the overall performance are still too slow). ### data-env-vars.name = mozilla-try-2023-03-22-ds2-pnm # benchmark.variants.explicit-rev = all-out-heads # benchmark.variants.issue6528 = disabled # benchmark.variants.protocol = ssh # benchmark.variants.pulled-delta-reuse-policy = default # bin-env-vars.hg.flavor = rust ## benchmark.variants.revs = any-1-extra-rev hg.command.pull: 6.517450 hg.command.push: 11.219888 ## benchmark.variants.revs = any-100-extra-rev hg.command.pull: 10.160991 hg.command.push: 14.251107 ### data-env-vars.name = mozilla-try-2023-03-22-zstd-sparse-revlog # bin-env-vars.hg.py-re2-module = default # benchmark.variants.explicit-rev = all-out-heads # benchmark.variants.issue6528 = disabled # benchmark.variants.protocol = ssh # benchmark.variants.pulled-delta-reuse-policy = default ## bin-env-vars.hg.flavor = default ## benchmark.variants.revs = any-1-extra-rev hg.command.pull: 8.577772 hg.command.push: 11.232059 ## bin-env-vars.hg.flavor = default ## benchmark.variants.revs = any-100-extra-rev hg.command.pull: 13.152976 hg.command.push: 15.367412 ## bin-env-vars.hg.flavor = rust ## benchmark.variants.revs = any-1-extra-rev hg.command.pull: 8.731982 hg.command.push: 11.178751 ## bin-env-vars.hg.flavor = rust ## benchmark.variants.revs = any-100-extra-rev hg.command.pull: 13.184236 hg.command.push: 15.620843

/*
 bdiff.c - efficient binary diff extension for Mercurial

 Copyright 2005, 2006 Olivia Mackall <olivia@selenic.com>

 This software may be used and distributed according to the terms of
 the GNU General Public License, incorporated herein by reference.

 Based roughly on Python difflib
*/

#define PY_SSIZE_T_CLEAN
#include <Python.h>
#include <limits.h>
#include <stdlib.h>
#include <string.h>

#include "bdiff.h"
#include "bitmanipulation.h"
#include "thirdparty/xdiff/xdiff.h"
#include "util.h"

static PyObject *blocks(PyObject *self, PyObject *args)
{
	PyObject *sa, *sb, *rl = NULL, *m;
	struct bdiff_line *a, *b;
	struct bdiff_hunk l, *h;
	int an, bn, count, pos = 0;

	l.next = NULL;

	if (!PyArg_ParseTuple(args, "SS:bdiff", &sa, &sb)) {
		return NULL;
	}

	an = bdiff_splitlines(PyBytes_AsString(sa), PyBytes_Size(sa), &a);
	bn = bdiff_splitlines(PyBytes_AsString(sb), PyBytes_Size(sb), &b);

	if (!a || !b) {
		goto nomem;
	}

	count = bdiff_diff(a, an, b, bn, &l);
	if (count < 0) {
		goto nomem;
	}

	rl = PyList_New(count);
	if (!rl) {
		goto nomem;
	}

	for (h = l.next; h; h = h->next) {
		m = Py_BuildValue("iiii", h->a1, h->a2, h->b1, h->b2);
		PyList_SetItem(rl, pos, m);
		pos++;
	}

nomem:
	free(a);
	free(b);
	bdiff_freehunks(l.next);
	return rl ? rl : PyErr_NoMemory();
}

static PyObject *bdiff(PyObject *self, PyObject *args)
{
	Py_buffer ba, bb;
	char *rb, *ia, *ib;
	PyObject *result = NULL;
	struct bdiff_line *al = NULL, *bl = NULL;
	struct bdiff_hunk l, *h;
	int an, bn, count;
	Py_ssize_t len = 0, la, lb, li = 0, lcommon = 0, lmax;
	PyThreadState *_save = NULL;

	l.next = NULL;

	if (!PyArg_ParseTuple(args, "y*y*:bdiff", &ba, &bb)) {
		return NULL;
	}

	if (!PyBuffer_IsContiguous(&ba, 'C') || ba.ndim > 1) {
		PyErr_SetString(PyExc_ValueError, "bdiff input not contiguous");
		goto cleanup;
	}

	if (!PyBuffer_IsContiguous(&bb, 'C') || bb.ndim > 1) {
		PyErr_SetString(PyExc_ValueError, "bdiff input not contiguous");
		goto cleanup;
	}

	la = ba.len;
	lb = bb.len;

	if (la > UINT_MAX || lb > UINT_MAX) {
		PyErr_SetString(PyExc_ValueError, "bdiff inputs too large");
		goto cleanup;
	}

	_save = PyEval_SaveThread();

	lmax = la > lb ? lb : la;
	for (ia = ba.buf, ib = bb.buf; li < lmax && *ia == *ib;
	     ++li, ++ia, ++ib) {
		if (*ia == '\n') {
			lcommon = li + 1;
		}
	}
	/* we can almost add: if (li == lmax) lcommon = li; */

	an = bdiff_splitlines((char *)ba.buf + lcommon, la - lcommon, &al);
	bn = bdiff_splitlines((char *)bb.buf + lcommon, lb - lcommon, &bl);
	if (!al || !bl) {
		PyErr_NoMemory();
		goto cleanup;
	}

	count = bdiff_diff(al, an, bl, bn, &l);
	if (count < 0) {
		PyErr_NoMemory();
		goto cleanup;
	}

	/* calculate length of output */
	la = lb = 0;
	for (h = l.next; h; h = h->next) {
		if (h->a1 != la || h->b1 != lb) {
			len += 12 + bl[h->b1].l - bl[lb].l;
		}
		la = h->a2;
		lb = h->b2;
	}
	PyEval_RestoreThread(_save);
	_save = NULL;

	result = PyBytes_FromStringAndSize(NULL, len);

	if (!result) {
		goto cleanup;
	}

	/* build binary patch */
	rb = PyBytes_AsString(result);
	la = lb = 0;

	for (h = l.next; h; h = h->next) {
		if (h->a1 != la || h->b1 != lb) {
			len = bl[h->b1].l - bl[lb].l;
			putbe32((uint32_t)(al[la].l + lcommon - al->l), rb);
			putbe32((uint32_t)(al[h->a1].l + lcommon - al->l),
			        rb + 4);
			putbe32((uint32_t)len, rb + 8);
			memcpy(rb + 12, bl[lb].l, len);
			rb += 12 + len;
		}
		la = h->a2;
		lb = h->b2;
	}

cleanup:
	if (_save) {
		PyEval_RestoreThread(_save);
	}
	PyBuffer_Release(&ba);
	PyBuffer_Release(&bb);
	free(al);
	free(bl);
	bdiff_freehunks(l.next);
	return result;
}

/*
 * If allws != 0, remove all whitespace (' ', \t and \r). Otherwise,
 * reduce whitespace sequences to a single space and trim remaining whitespace
 * from end of lines.
 */
static PyObject *fixws(PyObject *self, PyObject *args)
{
	PyObject *s, *result = NULL;
	char allws, c;
	const char *r;
	Py_ssize_t i, rlen, wlen = 0;
	char *w;

	if (!PyArg_ParseTuple(args, "Sb:fixws", &s, &allws)) {
		return NULL;
	}
	r = PyBytes_AsString(s);
	rlen = PyBytes_Size(s);

	w = (char *)PyMem_Malloc(rlen ? rlen : 1);
	if (!w) {
		goto nomem;
	}

	for (i = 0; i != rlen; i++) {
		c = r[i];
		if (c == ' ' || c == '\t' || c == '\r') {
			if (!allws && (wlen == 0 || w[wlen - 1] != ' ')) {
				w[wlen++] = ' ';
			}
		} else if (c == '\n' && !allws && wlen > 0 &&
		           w[wlen - 1] == ' ') {
			w[wlen - 1] = '\n';
		} else {
			w[wlen++] = c;
		}
	}

	result = PyBytes_FromStringAndSize(w, wlen);

nomem:
	PyMem_Free(w);
	return result ? result : PyErr_NoMemory();
}

static bool sliceintolist(PyObject *list, Py_ssize_t destidx,
                          const char *source, Py_ssize_t len)
{
	PyObject *sliced = PyBytes_FromStringAndSize(source, len);
	if (sliced == NULL) {
		return false;
	}
	PyList_SET_ITEM(list, destidx, sliced);
	return true;
}

static PyObject *splitnewlines(PyObject *self, PyObject *args)
{
	const char *text;
	Py_ssize_t nelts = 0, size, i, start = 0;
	PyObject *result = NULL;

	if (!PyArg_ParseTuple(args, "y#", &text, &size)) {
		goto abort;
	}
	if (!size) {
		return PyList_New(0);
	}
	/* This loops to size-1 because if the last byte is a newline,
	 * we don't want to perform a split there. */
	for (i = 0; i < size - 1; ++i) {
		if (text[i] == '\n') {
			++nelts;
		}
	}
	if ((result = PyList_New(nelts + 1)) == NULL) {
		goto abort;
	}
	nelts = 0;
	for (i = 0; i < size - 1; ++i) {
		if (text[i] == '\n') {
			if (!sliceintolist(result, nelts++, text + start,
			                   i - start + 1)) {
				goto abort;
			}
			start = i + 1;
		}
	}
	if (!sliceintolist(result, nelts++, text + start, size - start)) {
		goto abort;
	}
	return result;
abort:
	Py_XDECREF(result);
	return NULL;
}

static int hunk_consumer(int64_t a1, int64_t a2, int64_t b1, int64_t b2,
                         void *priv)
{
	PyObject *rl = (PyObject *)priv;
	PyObject *m = Py_BuildValue("LLLL", a1, a2, b1, b2);
	int r;
	if (!m) {
		return -1;
	}
	r = PyList_Append(rl, m);
	Py_DECREF(m);
	return r;
}

static PyObject *xdiffblocks(PyObject *self, PyObject *args)
{
	Py_ssize_t la, lb;
	mmfile_t a, b;
	PyObject *rl;

	xpparam_t xpp = {
	    XDF_INDENT_HEURISTIC, /* flags */
	};
	xdemitconf_t xecfg = {
	    XDL_EMIT_BDIFFHUNK, /* flags */
	    hunk_consumer,      /* hunk_consume_func */
	};
	xdemitcb_t ecb = {
	    NULL, /* priv */
	};

	if (!PyArg_ParseTuple(args, "y#y#", &a.ptr, &la, &b.ptr, &lb)) {
		return NULL;
	}

	a.size = la;
	b.size = lb;

	rl = PyList_New(0);
	if (!rl) {
		return PyErr_NoMemory();
	}

	ecb.priv = rl;

	if (xdl_diff(&a, &b, &xpp, &xecfg, &ecb) != 0) {
		Py_DECREF(rl);
		return PyErr_NoMemory();
	}

	return rl;
}

static char mdiff_doc[] = "Efficient binary diff.";

static PyMethodDef methods[] = {
    {"bdiff", bdiff, METH_VARARGS, "calculate a binary diff\n"},
    {"blocks", blocks, METH_VARARGS, "find a list of matching lines\n"},
    {"fixws", fixws, METH_VARARGS, "normalize diff whitespaces\n"},
    {"splitnewlines", splitnewlines, METH_VARARGS,
     "like str.splitlines, but only split on newlines\n"},
    {"xdiffblocks", xdiffblocks, METH_VARARGS,
     "find a list of matching lines using xdiff algorithm\n"},
    {NULL, NULL},
};

static const int version = 3;

static struct PyModuleDef bdiff_module = {
    PyModuleDef_HEAD_INIT, "bdiff", mdiff_doc, -1, methods,
};

PyMODINIT_FUNC PyInit_bdiff(void)
{
	PyObject *m;
	m = PyModule_Create(&bdiff_module);
	PyModule_AddIntConstant(m, "version", version);
	return m;
}