[apple/icu.git] / icuSources / tools / toolutil / writesrc.cpp

// © 2016 and later: Unicode, Inc. and others.
// License & terms of use: http://www.unicode.org/copyright.html
/*
*******************************************************************************
*
*   Copyright (C) 2005-2012, International Business Machines
*   Corporation and others.  All Rights Reserved.
*
*******************************************************************************
*   file name:  writesrc.c
*   encoding:   UTF-8
*   tab size:   8 (not used)
*   indentation:4
*
*   created on: 2005apr23
*   created by: Markus W. Scherer
*
*   Helper functions for writing source code for data.
*/

#include <stdio.h>
#include <time.h>
#include "unicode/utypes.h"
#include "unicode/putil.h"
#include "utrie2.h"
#include "cstring.h"
#include "writesrc.h"

static FILE *
usrc_createWithHeader(const char *path, const char *filename,
                      const char *generator, const char *header) {
    char buffer[1024];
    const char *p;
    char *q;
    FILE *f;
    char c;

    if(path==NULL) {
        p=filename;
    } else {
        /* concatenate path and filename, with U_FILE_SEP_CHAR in between if necessary */
        uprv_strcpy(buffer, path);
        q=buffer+uprv_strlen(buffer);
        if(q>buffer && (c=*(q-1))!=U_FILE_SEP_CHAR && c!=U_FILE_ALT_SEP_CHAR) {
            *q++=U_FILE_SEP_CHAR;
        }
        uprv_strcpy(q, filename);
        p=buffer;
    }

    f=fopen(p, "w");
    if(f!=NULL) {
        const struct tm *lt;
        time_t t;

        time(&t);
        lt=localtime(&t);
        if(generator==NULL) {
            strftime(buffer, sizeof(buffer), "%Y-%m-%d", lt);
            fprintf(f, header, filename, buffer);
        } else {
            fprintf(f, header, filename, generator);
        }
    } else {
        fprintf(
            stderr,
            "usrc_create(%s, %s): unable to create file\n",
            path!=NULL ? path : "", filename);
    }
    return f;
}

U_CAPI FILE * U_EXPORT2
usrc_create(const char *path, const char *filename, const char *generator) {
    // TODO: Add parameter for the first year this file was generated, not before 2016.
    static const char *header=
        "// © 2016 and later: Unicode, Inc. and others.\n"
        "// License & terms of use: http://www.unicode.org/copyright.html\n"
        "//\n"
        "// Copyright (C) 1999-2016, International Business Machines\n"
        "// Corporation and others.  All Rights Reserved.\n"
        "//\n"
        "// file name: %s\n"
        "//\n"
        "// machine-generated by: %s\n"
        "\n\n";
    return usrc_createWithHeader(path, filename, generator, header);
}

U_CAPI FILE * U_EXPORT2
usrc_createTextData(const char *path, const char *filename, const char *generator) {
    // TODO: Add parameter for the first year this file was generated, not before 2016.
    static const char *header=
        "# Copyright (C) 2016 and later: Unicode, Inc. and others.\n"
        "# License & terms of use: http://www.unicode.org/copyright.html\n"
        "# Copyright (C) 1999-2016, International Business Machines\n"
        "# Corporation and others.  All Rights Reserved.\n"
        "#\n"
        "# file name: %s\n"
        "#\n"
        "# machine-generated by: %s\n"
        "\n\n";
    return usrc_createWithHeader(path, filename, generator, header);
}

U_CAPI void U_EXPORT2
usrc_writeArray(FILE *f,
                const char *prefix,
                const void *p, int32_t width, int32_t length,
                const char *postfix) {
    const uint8_t *p8;
    const uint16_t *p16;
    const uint32_t *p32;
    uint32_t value;
    int32_t i, col;

    p8=NULL;
    p16=NULL;
    p32=NULL;
    switch(width) {
    case 8:
        p8=(const uint8_t *)p;
        break;
    case 16:
        p16=(const uint16_t *)p;
        break;
    case 32:
        p32=(const uint32_t *)p;
        break;
    default:
        fprintf(stderr, "usrc_writeArray(width=%ld) unrecognized width\n", (long)width);
        return;
    }
    if(prefix!=NULL) {
        fprintf(f, prefix, (long)length);
    }
    for(i=col=0; i<length; ++i, ++col) {
        if(i>0) {
            if(col<16) {
                fputc(',', f);
            } else {
                fputs(",\n", f);
                col=0;
            }
        }
        switch(width) {
        case 8:
            value=p8[i];
            break;
        case 16:
            value=p16[i];
            break;
        case 32:
            value=p32[i];
            break;
        default:
            value=0; /* unreachable */
            break;
        }
        fprintf(f, value<=9 ? "%lu" : "0x%lx", (unsigned long)value);
    }
    if(postfix!=NULL) {
        fputs(postfix, f);
    }
}

U_CAPI void U_EXPORT2
usrc_writeUTrie2Arrays(FILE *f,
                       const char *indexPrefix, const char *data32Prefix,
                       const UTrie2 *pTrie,
                       const char *postfix) {
    if(pTrie->data32==NULL) {
        /* 16-bit trie */
        usrc_writeArray(f, indexPrefix, pTrie->index, 16, pTrie->indexLength+pTrie->dataLength, postfix);
    } else {
        /* 32-bit trie */
        usrc_writeArray(f, indexPrefix, pTrie->index, 16, pTrie->indexLength, postfix);
        usrc_writeArray(f, data32Prefix, pTrie->data32, 32, pTrie->dataLength, postfix);
    }
}

U_CAPI void U_EXPORT2
usrc_writeUTrie2Struct(FILE *f,
                       const char *prefix,
                       const UTrie2 *pTrie,
                       const char *indexName, const char *data32Name,
                       const char *postfix) {
    if(prefix!=NULL) {
        fputs(prefix, f);
    }
    if(pTrie->data32==NULL) {
        /* 16-bit trie */
        fprintf(
            f,
            "    %s,\n"         /* index */
            "    %s+%ld,\n"     /* data16 */
            "    NULL,\n",      /* data32 */
            indexName,
            indexName, 
            (long)pTrie->indexLength);
    } else {
        /* 32-bit trie */
        fprintf(
            f,
            "    %s,\n"         /* index */
            "    NULL,\n"       /* data16 */
            "    %s,\n",        /* data32 */
            indexName,
            data32Name);
    }
    fprintf(
        f,
        "    %ld,\n"            /* indexLength */
        "    %ld,\n"            /* dataLength */
        "    0x%hx,\n"          /* index2NullOffset */
        "    0x%hx,\n"          /* dataNullOffset */
        "    0x%lx,\n"          /* initialValue */
        "    0x%lx,\n"          /* errorValue */
        "    0x%lx,\n"          /* highStart */
        "    0x%lx,\n"          /* highValueIndex */
        "    NULL, 0, FALSE, FALSE, 0, NULL\n",
        (long)pTrie->indexLength, (long)pTrie->dataLength,
        (short)pTrie->index2NullOffset, (short)pTrie->dataNullOffset,
        (long)pTrie->initialValue, (long)pTrie->errorValue,
        (long)pTrie->highStart, (long)pTrie->highValueIndex);
    if(postfix!=NULL) {
        fputs(postfix, f);
    }
}

U_CAPI void U_EXPORT2
usrc_writeArrayOfMostlyInvChars(FILE *f,
                                const char *prefix,
                                const char *p, int32_t length,
                                const char *postfix) {
    int32_t i, col;
    int prev2, prev, c;

    if(prefix!=NULL) {
        fprintf(f, prefix, (long)length);
    }
    prev2=prev=-1;
    for(i=col=0; i<length; ++i, ++col) {
        c=(uint8_t)p[i];
        if(i>0) {
            /* Break long lines. Try to break at interesting places, to minimize revision diffs. */
            if( 
                /* Very long line. */
                col>=32 ||
                /* Long line, break after terminating NUL. */
                (col>=24 && prev2>=0x20 && prev==0) ||
                /* Medium-long line, break before non-NUL, non-character byte. */
                (col>=16 && (prev==0 || prev>=0x20) && 0<c && c<0x20)
            ) {
                fputs(",\n", f);
                col=0;
            } else {
                fputc(',', f);
            }
        }
        fprintf(f, c<0x20 ? "%u" : "'%c'", c);
        prev2=prev;
        prev=c;
    }
    if(postfix!=NULL) {
        fputs(postfix, f);
    }
}
Commit	Line	Data
f3c0d7a5 A	1	// © 2016 and later: Unicode, Inc. and others.
f3c0d7a5 A	2	// License & terms of use: http://www.unicode.org/copyright.html
73c04bcf A	3	/*
	4	*******************************************************************************
	5	*
4388f060	6	* Copyright (C) 2005-2012, International Business Machines
73c04bcf A	7	* Corporation and others. All Rights Reserved.
	8	*
	9	*******************************************************************************
	10	* file name: writesrc.c
f3c0d7a5	11	* encoding: UTF-8
73c04bcf A	12	* tab size: 8 (not used)
	13	* indentation:4
	14	*
	15	* created on: 2005apr23
	16	* created by: Markus W. Scherer
	17	*
	18	* Helper functions for writing source code for data.
	19	*/
	20
	21	#include <stdio.h>
	22	#include <time.h>
	23	#include "unicode/utypes.h"
	24	#include "unicode/putil.h"
729e4ab9	25	#include "utrie2.h"
73c04bcf A	26	#include "cstring.h"
	27	#include "writesrc.h"
	28
729e4ab9	29	static FILE *
4388f060 A	30	usrc_createWithHeader(const char path, const char filename,
4388f060 A	31	const char generator, const char header) {
73c04bcf A	32	char buffer[1024];
	33	const char *p;
	34	char *q;
	35	FILE *f;
	36	char c;
	37
	38	if(path==NULL) {
	39	p=filename;
	40	} else {
	41	/* concatenate path and filename, with U_FILE_SEP_CHAR in between if necessary */
	42	uprv_strcpy(buffer, path);
	43	q=buffer+uprv_strlen(buffer);
	44	if(q>buffer && (c=*(q-1))!=U_FILE_SEP_CHAR && c!=U_FILE_ALT_SEP_CHAR) {
	45	*q++=U_FILE_SEP_CHAR;
	46	}
	47	uprv_strcpy(q, filename);
	48	p=buffer;
	49	}
	50
	51	f=fopen(p, "w");
	52	if(f!=NULL) {
73c04bcf A	53	const struct tm *lt;
	54	time_t t;
	55
	56	time(&t);
	57	lt=localtime(&t);
4388f060 A	58	if(generator==NULL) {
4388f060 A	59	strftime(buffer, sizeof(buffer), "%Y-%m-%d", lt);
f3c0d7a5	60	fprintf(f, header, filename, buffer);
4388f060	61	} else {
f3c0d7a5	62	fprintf(f, header, filename, generator);
4388f060	63	}
73c04bcf A	64	} else {
	65	fprintf(
	66	stderr,
	67	"usrc_create(%s, %s): unable to create file\n",
46f4442e	68	path!=NULL ? path : "", filename);
73c04bcf A	69	}
	70	return f;
	71	}
	72
729e4ab9	73	U_CAPI FILE * U_EXPORT2
4388f060	74	usrc_create(const char path, const char filename, const char *generator) {
f3c0d7a5	75	// TODO: Add parameter for the first year this file was generated, not before 2016.
4388f060	76	static const char *header=
f3c0d7a5 A	77	"// © 2016 and later: Unicode, Inc. and others.\n"
	78	"// License & terms of use: http://www.unicode.org/copyright.html\n"
	79	"//\n"
	80	"// Copyright (C) 1999-2016, International Business Machines\n"
	81	"// Corporation and others. All Rights Reserved.\n"
	82	"//\n"
	83	"// file name: %s\n"
	84	"//\n"
	85	"// machine-generated by: %s\n"
	86	"\n\n";
4388f060	87	return usrc_createWithHeader(path, filename, generator, header);
729e4ab9 A	88	}
	89
	90	U_CAPI FILE * U_EXPORT2
4388f060	91	usrc_createTextData(const char path, const char filename, const char *generator) {
f3c0d7a5	92	// TODO: Add parameter for the first year this file was generated, not before 2016.
4388f060	93	static const char *header=
f3c0d7a5 A	94	"# Copyright (C) 2016 and later: Unicode, Inc. and others.\n"
	95	"# License & terms of use: http://www.unicode.org/copyright.html\n"
	96	"# Copyright (C) 1999-2016, International Business Machines\n"
729e4ab9 A	97	"# Corporation and others. All Rights Reserved.\n"
	98	"#\n"
	99	"# file name: %s\n"
	100	"#\n"
4388f060	101	"# machine-generated by: %s\n"
f3c0d7a5	102	"\n\n";
4388f060	103	return usrc_createWithHeader(path, filename, generator, header);
729e4ab9 A	104	}
729e4ab9 A	105
73c04bcf A	106	U_CAPI void U_EXPORT2
	107	usrc_writeArray(FILE *f,
	108	const char *prefix,
	109	const void *p, int32_t width, int32_t length,
	110	const char *postfix) {
	111	const uint8_t *p8;
	112	const uint16_t *p16;
	113	const uint32_t *p32;
	114	uint32_t value;
	115	int32_t i, col;
	116
	117	p8=NULL;
	118	p16=NULL;
	119	p32=NULL;
	120	switch(width) {
	121	case 8:
	122	p8=(const uint8_t *)p;
	123	break;
	124	case 16:
	125	p16=(const uint16_t *)p;
	126	break;
	127	case 32:
	128	p32=(const uint32_t *)p;
	129	break;
	130	default:
	131	fprintf(stderr, "usrc_writeArray(width=%ld) unrecognized width\n", (long)width);
	132	return;
	133	}
	134	if(prefix!=NULL) {
	135	fprintf(f, prefix, (long)length);
	136	}
	137	for(i=col=0; i<length; ++i, ++col) {
	138	if(i>0) {
	139	if(col<16) {
	140	fputc(',', f);
	141	} else {
	142	fputs(",\n", f);
	143	col=0;
	144	}
	145	}
	146	switch(width) {
	147	case 8:
	148	value=p8[i];
	149	break;
	150	case 16:
	151	value=p16[i];
	152	break;
	153	case 32:
	154	value=p32[i];
	155	break;
	156	default:
	157	value=0; /* unreachable */
	158	break;
	159	}
	160	fprintf(f, value<=9 ? "%lu" : "0x%lx", (unsigned long)value);
	161	}
	162	if(postfix!=NULL) {
	163	fputs(postfix, f);
	164	}
	165	}
	166
	167	U_CAPI void U_EXPORT2
729e4ab9 A	168	usrc_writeUTrie2Arrays(FILE *f,
	169	const char indexPrefix, const char data32Prefix,
	170	const UTrie2 *pTrie,
	171	const char *postfix) {
73c04bcf A	172	if(pTrie->data32==NULL) {
	173	/* 16-bit trie */
	174	usrc_writeArray(f, indexPrefix, pTrie->index, 16, pTrie->indexLength+pTrie->dataLength, postfix);
	175	} else {
	176	/* 32-bit trie */
	177	usrc_writeArray(f, indexPrefix, pTrie->index, 16, pTrie->indexLength, postfix);
729e4ab9	178	usrc_writeArray(f, data32Prefix, pTrie->data32, 32, pTrie->dataLength, postfix);
73c04bcf A	179	}
	180	}
	181
	182	U_CAPI void U_EXPORT2
729e4ab9 A	183	usrc_writeUTrie2Struct(FILE *f,
	184	const char *prefix,
	185	const UTrie2 *pTrie,
	186	const char indexName, const char data32Name,
	187	const char *postfix) {
73c04bcf A	188	if(prefix!=NULL) {
	189	fputs(prefix, f);
	190	}
729e4ab9 A	191	if(pTrie->data32==NULL) {
	192	/* 16-bit trie */
	193	fprintf(
	194	f,
	195	" %s,\n" /* index */
	196	" %s+%ld,\n" /* data16 */
	197	" NULL,\n", /* data32 */
	198	indexName,
	199	indexName,
	200	(long)pTrie->indexLength);
	201	} else {
	202	/* 32-bit trie */
	203	fprintf(
	204	f,
	205	" %s,\n" /* index */
	206	" NULL,\n" /* data16 */
	207	" %s,\n", /* data32 */
	208	indexName,
	209	data32Name);
73c04bcf A	210	}
	211	fprintf(
	212	f,
729e4ab9 A	213	" %ld,\n" /* indexLength */
	214	" %ld,\n" /* dataLength */
	215	" 0x%hx,\n" /* index2NullOffset */
	216	" 0x%hx,\n" /* dataNullOffset */
	217	" 0x%lx,\n" /* initialValue */
	218	" 0x%lx,\n" /* errorValue */
	219	" 0x%lx,\n" /* highStart */
	220	" 0x%lx,\n" /* highValueIndex */
	221	" NULL, 0, FALSE, FALSE, 0, NULL\n",
73c04bcf	222	(long)pTrie->indexLength, (long)pTrie->dataLength,
729e4ab9 A	223	(short)pTrie->index2NullOffset, (short)pTrie->dataNullOffset,
	224	(long)pTrie->initialValue, (long)pTrie->errorValue,
	225	(long)pTrie->highStart, (long)pTrie->highValueIndex);
73c04bcf A	226	if(postfix!=NULL) {
	227	fputs(postfix, f);
	228	}
	229	}
4388f060 A	230
	231	U_CAPI void U_EXPORT2
	232	usrc_writeArrayOfMostlyInvChars(FILE *f,
	233	const char *prefix,
	234	const char *p, int32_t length,
	235	const char *postfix) {
	236	int32_t i, col;
	237	int prev2, prev, c;
	238
	239	if(prefix!=NULL) {
	240	fprintf(f, prefix, (long)length);
	241	}
	242	prev2=prev=-1;
	243	for(i=col=0; i<length; ++i, ++col) {
	244	c=(uint8_t)p[i];
	245	if(i>0) {
	246	/* Break long lines. Try to break at interesting places, to minimize revision diffs. */
	247	if(
	248	/* Very long line. */
	249	col>=32 \|\|
	250	/* Long line, break after terminating NUL. */
	251	(col>=24 && prev2>=0x20 && prev==0) \|\|
	252	/* Medium-long line, break before non-NUL, non-character byte. */
	253	(col>=16 && (prev==0 \|\| prev>=0x20) && 0<c && c<0x20)
	254	) {
	255	fputs(",\n", f);
	256	col=0;
	257	} else {
	258	fputc(',', f);
	259	}
	260	}
	261	fprintf(f, c<0x20 ? "%u" : "'%c'", c);
	262	prev2=prev;
	263	prev=c;
	264	}
	265	if(postfix!=NULL) {
	266	fputs(postfix, f);
	267	}
	268	}