X-Git-Url: https://mattmccutchen.net/rsync/rsync.git/blobdiff_plain/06e27ef78e0348ec7326f1c1c3c1026abac9db42..d4d56eed8a2dfdd2cd78a19f9c1bae02496427e3:/zlib/trees.c diff --git a/zlib/trees.c b/zlib/trees.c index 41b7fe8f..395e4e16 100644 --- a/zlib/trees.c +++ b/zlib/trees.c @@ -1,6 +1,6 @@ /* trees.c -- output deflated data using Huffman coding - * Copyright (C) 1995-1998 Jean-loup Gailly - * For conditions of distribution and use, see copyright notice in zlib.h + * Copyright (C) 1995-2005 Jean-loup Gailly + * For conditions of distribution and use, see copyright notice in zlib.h */ /* @@ -230,9 +230,6 @@ local void send_bits(s, value, length) #endif /* DEBUG */ -#ifndef MAX -#define MAX(a,b) ((a) >= (b) ? (a) : (b)) -#endif /* the arguments must not have side effects */ /* =========================================================================== @@ -252,6 +249,13 @@ local void tr_static_init() if (static_init_done) return; + /* For some embedded targets, global variables are not initialized: */ + static_l_desc.static_tree = static_ltree; + static_l_desc.extra_bits = extra_lbits; + static_d_desc.static_tree = static_dtree; + static_d_desc.extra_bits = extra_dbits; + static_bl_desc.extra_bits = extra_blbits; + /* Initialize the mapping length (0..255) -> length code (0..28) */ length = 0; for (code = 0; code < LENGTH_CODES-1; code++) { @@ -330,42 +334,42 @@ void gen_trees_header() Assert (header != NULL, "Can't open trees.h"); fprintf(header, - "/* header created automatically with -DGEN_TREES_H */\n\n"); + "/* header created automatically with -DGEN_TREES_H */\n\n"); fprintf(header, "local const ct_data static_ltree[L_CODES+2] = {\n"); for (i = 0; i < L_CODES+2; i++) { - fprintf(header, "{{%3u},{%3u}}%s", static_ltree[i].Code, - static_ltree[i].Len, SEPARATOR(i, L_CODES+1, 5)); + fprintf(header, "{{%3u},{%3u}}%s", static_ltree[i].Code, + static_ltree[i].Len, SEPARATOR(i, L_CODES+1, 5)); } fprintf(header, "local const ct_data static_dtree[D_CODES] = {\n"); for (i = 0; i < D_CODES; i++) { - fprintf(header, "{{%2u},{%2u}}%s", static_dtree[i].Code, - static_dtree[i].Len, SEPARATOR(i, D_CODES-1, 5)); + fprintf(header, "{{%2u},{%2u}}%s", static_dtree[i].Code, + static_dtree[i].Len, SEPARATOR(i, D_CODES-1, 5)); } fprintf(header, "const uch _dist_code[DIST_CODE_LEN] = {\n"); for (i = 0; i < DIST_CODE_LEN; i++) { - fprintf(header, "%2u%s", _dist_code[i], - SEPARATOR(i, DIST_CODE_LEN-1, 20)); + fprintf(header, "%2u%s", _dist_code[i], + SEPARATOR(i, DIST_CODE_LEN-1, 20)); } fprintf(header, "const uch _length_code[MAX_MATCH-MIN_MATCH+1]= {\n"); for (i = 0; i < MAX_MATCH-MIN_MATCH+1; i++) { - fprintf(header, "%2u%s", _length_code[i], - SEPARATOR(i, MAX_MATCH-MIN_MATCH, 20)); + fprintf(header, "%2u%s", _length_code[i], + SEPARATOR(i, MAX_MATCH-MIN_MATCH, 20)); } fprintf(header, "local const int base_length[LENGTH_CODES] = {\n"); for (i = 0; i < LENGTH_CODES; i++) { - fprintf(header, "%1u%s", base_length[i], - SEPARATOR(i, LENGTH_CODES-1, 20)); + fprintf(header, "%1u%s", base_length[i], + SEPARATOR(i, LENGTH_CODES-1, 20)); } fprintf(header, "local const int base_dist[D_CODES] = {\n"); for (i = 0; i < D_CODES; i++) { - fprintf(header, "%5u%s", base_dist[i], - SEPARATOR(i, D_CODES-1, 10)); + fprintf(header, "%5u%s", base_dist[i], + SEPARATOR(i, D_CODES-1, 10)); } fclose(header); @@ -380,8 +384,6 @@ void _tr_init(s) { tr_static_init(); - s->compressed_len = 0L; - s->l_desc.dyn_tree = s->dyn_ltree; s->l_desc.stat_desc = &static_l_desc; @@ -395,6 +397,7 @@ void _tr_init(s) s->bi_valid = 0; s->last_eob_len = 8; /* enough lookahead for inflate */ #ifdef DEBUG + s->compressed_len = 0L; s->bits_sent = 0L; #endif @@ -499,7 +502,7 @@ local void gen_bitlen(s, desc) int bits; /* bit length */ int xbits; /* extra bits */ ush f; /* frequency */ - int Overflow = 0; /* number of elements with bit length too large */ + int overflow = 0; /* number of elements with bit length too large */ for (bits = 0; bits <= MAX_BITS; bits++) s->bl_count[bits] = 0; @@ -511,7 +514,7 @@ local void gen_bitlen(s, desc) for (h = s->heap_max+1; h < HEAP_SIZE; h++) { n = s->heap[h]; bits = tree[tree[n].Dad].Len + 1; - if (bits > max_length) bits = max_length, Overflow++; + if (bits > max_length) bits = max_length, overflow++; tree[n].Len = (ush)bits; /* We overwrite tree[n].Dad which is no longer needed */ @@ -524,7 +527,7 @@ local void gen_bitlen(s, desc) s->opt_len += (ulg)f * (bits + xbits); if (stree) s->static_len += (ulg)f * (stree[n].Len + xbits); } - if (Overflow == 0) return; + if (overflow == 0) return; Trace((stderr,"\nbit length overflow\n")); /* This happens for example on obj2 and pic of the Calgary corpus */ @@ -539,8 +542,8 @@ local void gen_bitlen(s, desc) /* The brother of the overflow item also moves one step up, * but this does not affect bl_count[max_length] */ - Overflow -= 2; - } while (Overflow > 0); + overflow -= 2; + } while (overflow > 0); /* Now recompute all bit lengths, scanning in increasing frequency. * h is still equal to HEAP_SIZE. (It is simpler to reconstruct all @@ -552,7 +555,7 @@ local void gen_bitlen(s, desc) while (n != 0) { m = s->heap[--h]; if (m > max_code) continue; - if (tree[m].Len != (unsigned) bits) { + if ((unsigned) tree[m].Len != (unsigned) bits) { Trace((stderr,"code %d bits %d->%d\n", m, tree[m].Len, bits)); s->opt_len += ((long)bits - (long)tree[m].Len) *(long)tree[m].Freq; @@ -671,7 +674,8 @@ local void build_tree(s, desc) /* Create a new node father of n and m */ tree[node].Freq = tree[n].Freq + tree[m].Freq; - s->depth[node] = (uch) (MAX(s->depth[n], s->depth[m]) + 1); + s->depth[node] = (uch)((s->depth[n] >= s->depth[m] ? + s->depth[n] : s->depth[m]) + 1); tree[n].Dad = tree[m].Dad = (ush)node; #ifdef DUMP_BL_TREE if (tree == s->bl_tree) { @@ -867,9 +871,10 @@ void _tr_stored_block(s, buf, stored_len, eof) int eof; /* true if this is the last block for a file */ { send_bits(s, (STORED_BLOCK<<1)+eof, 3); /* send block type */ +#ifdef DEBUG s->compressed_len = (s->compressed_len + 3 + 7) & (ulg)~7L; s->compressed_len += (stored_len + 4) << 3; - +#endif copy_block(s, buf, (unsigned)stored_len, 1); /* with header */ } @@ -889,7 +894,9 @@ void _tr_align(s) { send_bits(s, STATIC_TREES<<1, 3); send_code(s, END_BLOCK, static_ltree); +#ifdef DEBUG s->compressed_len += 10L; /* 3 for block type, 7 for EOB */ +#endif bi_flush(s); /* Of the 10 bits for the empty block, we have already sent * (10 - bi_valid) bits. The lookahead for the last real code (before @@ -899,7 +906,9 @@ void _tr_align(s) if (1 + s->last_eob_len + 10 - s->bi_valid < 9) { send_bits(s, STATIC_TREES<<1, 3); send_code(s, END_BLOCK, static_ltree); +#ifdef DEBUG s->compressed_len += 10L; +#endif bi_flush(s); } s->last_eob_len = 7; @@ -907,10 +916,9 @@ void _tr_align(s) /* =========================================================================== * Determine the best encoding for the current block: dynamic trees, static - * trees or store, and output the encoded block to the zip file. This function - * returns the total compressed length for the file so far. + * trees or store, and output the encoded block to the zip file. */ -ulg _tr_flush_block(s, buf, stored_len, eof) +void _tr_flush_block(s, buf, stored_len, eof) deflate_state *s; charf *buf; /* input block, or NULL if too old */ ulg stored_len; /* length of input block */ @@ -922,60 +930,42 @@ ulg _tr_flush_block(s, buf, stored_len, eof) /* Build the Huffman trees unless a stored block is forced */ if (s->level > 0) { - /* Check if the file is ascii or binary */ - if (s->data_type == Z_UNKNOWN) set_data_type(s); + /* Check if the file is binary or text */ + if (stored_len > 0 && s->strm->data_type == Z_UNKNOWN) + set_data_type(s); - /* Construct the literal and distance trees */ - build_tree(s, (tree_desc *)(&(s->l_desc))); - Tracev((stderr, "\nlit data: dyn %ld, stat %ld", s->opt_len, - s->static_len)); + /* Construct the literal and distance trees */ + build_tree(s, (tree_desc *)(&(s->l_desc))); + Tracev((stderr, "\nlit data: dyn %ld, stat %ld", s->opt_len, + s->static_len)); - build_tree(s, (tree_desc *)(&(s->d_desc))); - Tracev((stderr, "\ndist data: dyn %ld, stat %ld", s->opt_len, - s->static_len)); - /* At this point, opt_len and static_len are the total bit lengths of - * the compressed block data, excluding the tree representations. - */ + build_tree(s, (tree_desc *)(&(s->d_desc))); + Tracev((stderr, "\ndist data: dyn %ld, stat %ld", s->opt_len, + s->static_len)); + /* At this point, opt_len and static_len are the total bit lengths of + * the compressed block data, excluding the tree representations. + */ - /* Build the bit length tree for the above two trees, and get the index - * in bl_order of the last bit length code to send. - */ - max_blindex = build_bl_tree(s); + /* Build the bit length tree for the above two trees, and get the index + * in bl_order of the last bit length code to send. + */ + max_blindex = build_bl_tree(s); - /* Determine the best encoding. Compute first the block length in bytes*/ - opt_lenb = (s->opt_len+3+7)>>3; - static_lenb = (s->static_len+3+7)>>3; + /* Determine the best encoding. Compute the block lengths in bytes. */ + opt_lenb = (s->opt_len+3+7)>>3; + static_lenb = (s->static_len+3+7)>>3; - Tracev((stderr, "\nopt %lu(%lu) stat %lu(%lu) stored %lu lit %u ", - opt_lenb, s->opt_len, static_lenb, s->static_len, stored_len, - s->last_lit)); + Tracev((stderr, "\nopt %lu(%lu) stat %lu(%lu) stored %lu lit %u ", + opt_lenb, s->opt_len, static_lenb, s->static_len, stored_len, + s->last_lit)); - if (static_lenb <= opt_lenb) opt_lenb = static_lenb; + if (static_lenb <= opt_lenb) opt_lenb = static_lenb; } else { Assert(buf != (char*)0, "lost buf"); - opt_lenb = static_lenb = stored_len + 5; /* force a stored block */ + opt_lenb = static_lenb = stored_len + 5; /* force a stored block */ } - /* If compression failed and this is the first and last block, - * and if the .zip file can be seeked (to rewrite the local header), - * the whole file is transformed into a stored file: - */ -#ifdef STORED_FILE_OK -# ifdef FORCE_STORED_FILE - if (eof && s->compressed_len == 0L) { /* force stored file */ -# else - if (stored_len <= opt_lenb && eof && s->compressed_len==0L && seekable()) { -# endif - /* Since LIT_BUFSIZE <= 2*WSIZE, the input data must be there: */ - if (buf == (charf*)0) error ("block vanished"); - - copy_block(buf, (unsigned)stored_len, 0); /* without header */ - s->compressed_len = stored_len << 3; - s->method = STORED; - } else -#endif /* STORED_FILE_OK */ - #ifdef FORCE_STORED if (buf != (char*)0) { /* force stored block */ #else @@ -993,29 +983,36 @@ ulg _tr_flush_block(s, buf, stored_len, eof) #ifdef FORCE_STATIC } else if (static_lenb >= 0) { /* force static trees */ #else - } else if (static_lenb == opt_lenb) { + } else if (s->strategy == Z_FIXED || static_lenb == opt_lenb) { #endif send_bits(s, (STATIC_TREES<<1)+eof, 3); compress_block(s, (ct_data *)static_ltree, (ct_data *)static_dtree); +#ifdef DEBUG s->compressed_len += 3 + s->static_len; +#endif } else { send_bits(s, (DYN_TREES<<1)+eof, 3); send_all_trees(s, s->l_desc.max_code+1, s->d_desc.max_code+1, max_blindex+1); compress_block(s, (ct_data *)s->dyn_ltree, (ct_data *)s->dyn_dtree); +#ifdef DEBUG s->compressed_len += 3 + s->opt_len; +#endif } Assert (s->compressed_len == s->bits_sent, "bad compressed size"); + /* The above check is made mod 2^32, for files larger than 512 MB + * and uLong implemented on 32 bits. + */ init_block(s); if (eof) { bi_windup(s); +#ifdef DEBUG s->compressed_len += 7; /* align on byte boundary */ +#endif } Tracev((stderr,"\ncomprlen %lu(%lu) ", s->compressed_len>>3, s->compressed_len-7*eof)); - - return s->compressed_len >> 3; } /* =========================================================================== @@ -1111,7 +1108,8 @@ local void compress_block(s, ltree, dtree) } /* literal or match pair ? */ /* Check that the overlay between pending_buf and d_buf+l_buf is ok: */ - Assert(s->pending < s->lit_bufsize + 2*lx, "pendingBuf overflow"); + Assert((uInt)(s->pending) < s->lit_bufsize + 2*lx, + "pendingBuf overflow"); } while (lx < s->last_lit); @@ -1120,21 +1118,24 @@ local void compress_block(s, ltree, dtree) } /* =========================================================================== - * Set the data type to ASCII or BINARY, using a crude approximation: - * binary if more than 20% of the bytes are <= 6 or >= 128, ascii otherwise. - * IN assertion: the fields freq of dyn_ltree are set and the total of all - * frequencies does not exceed 64K (to fit in an int on 16 bit machines). + * Set the data type to BINARY or TEXT, using a crude approximation: + * set it to Z_TEXT if all symbols are either printable characters (33 to 255) + * or white spaces (9 to 13, or 32); or set it to Z_BINARY otherwise. + * IN assertion: the fields Freq of dyn_ltree are set. */ local void set_data_type(s) deflate_state *s; { - int n = 0; - unsigned ascii_freq = 0; - unsigned bin_freq = 0; - while (n < 7) bin_freq += s->dyn_ltree[n++].Freq; - while (n < 128) ascii_freq += s->dyn_ltree[n++].Freq; - while (n < LITERALS) bin_freq += s->dyn_ltree[n++].Freq; - s->data_type = (Byte)(bin_freq > (ascii_freq >> 2) ? Z_BINARY : Z_ASCII); + int n; + + for (n = 0; n < 9; n++) + if (s->dyn_ltree[n].Freq != 0) + break; + if (n == 9) + for (n = 14; n < 32; n++) + if (s->dyn_ltree[n].Freq != 0) + break; + s->strm->data_type = (n == 32) ? Z_TEXT : Z_BINARY; } /* =========================================================================== @@ -1203,7 +1204,7 @@ local void copy_block(s, buf, len, header) s->last_eob_len = 8; /* enough lookahead for inflate */ if (header) { - put_short(s, (ush)len); + put_short(s, (ush)len); put_short(s, (ush)~len); #ifdef DEBUG s->bits_sent += 2*16;