|
65
|
1 /*
|
|
|
2 fcgo.go: Fucho de Go
|
|
|
3
|
|
|
4 Last Change: 2021-06-14 月 16:14:04.
|
|
|
5 */
|
|
|
6 package main
|
|
|
7
|
|
|
8 /*
|
|
|
9 #cgo LDFLAGS: -L. -lxdwapi -static
|
|
|
10 #include <stdio.h>
|
|
|
11 #include <stdlib.h>
|
|
|
12 #include <string.h>
|
|
|
13 #include <io.h>
|
|
|
14 #include <windows.h>
|
|
|
15 #include <xdw_api.h>
|
|
|
16 #include <xdwapian.h>
|
|
|
17
|
|
|
18 #define MAXLINE 12000
|
|
|
19 #define BLOCKSZ 128
|
|
|
20
|
|
|
21 void xdw2txt(const char* xdwfile, const char* txtfile) {
|
|
|
22 char in_path[_MAX_PATH];
|
|
|
23 _fullpath(in_path, xdwfile, _MAX_PATH);
|
|
|
24
|
|
|
25 XDW_DOCUMENT_HANDLE h = NULL;
|
|
|
26 XDW_OPEN_MODE_EX mode = {sizeof(XDW_OPEN_MODE_EX), XDW_OPEN_READONLY, XDW_AUTH_NODIALOGUE};
|
|
|
27 if (XDW_OpenDocumentHandle(in_path, &h, (XDW_OPEN_MODE*)&mode)) {
|
|
|
28 printf("XDW Error: cannot open %s\n", xdwfile);
|
|
|
29 return;
|
|
|
30 }
|
|
|
31
|
|
|
32 int api_result = XDW_GetFullText(h, txtfile, NULL);
|
|
|
33 if (api_result < 0) {
|
|
|
34 printf("Error: cannot write text\n");
|
|
|
35 return;
|
|
|
36 }
|
|
|
37 XDW_CloseDocumentHandle(h, NULL);
|
|
|
38 }
|
|
|
39
|
|
|
40 void xdwsplit1(const char* xdwfile) {
|
|
|
41 char in_path[_MAX_PATH];
|
|
|
42 _fullpath(in_path, xdwfile, _MAX_PATH);
|
|
|
43
|
|
|
44 XDW_DOCUMENT_HANDLE h = NULL;
|
|
|
45 XDW_OPEN_MODE_EX mode = {sizeof(XDW_OPEN_MODE_EX), XDW_OPEN_READONLY, XDW_AUTH_NODIALOGUE};
|
|
|
46 if (XDW_OpenDocumentHandle(in_path, &h, (XDW_OPEN_MODE*)&mode)) {
|
|
|
47 printf("Error: cannot open %s\n", xdwfile);
|
|
|
48 return;
|
|
|
49 }
|
|
|
50
|
|
|
51 XDW_DOCUMENT_INFO info = {sizeof(XDW_DOCUMENT_INFO), 0, 0, 0};
|
|
|
52 XDW_GetDocumentInformation(h, &info);
|
|
|
53 int nPage = info.nPages;
|
|
|
54
|
|
|
55 char buf[_MAX_PATH];
|
|
|
56 for (int i = 1; i <= nPage; i++) {
|
|
|
57 sprintf(buf, "fctmp_%05d.xdw", i);
|
|
|
58 _fullpath(in_path, buf, _MAX_PATH);
|
|
|
59
|
|
|
60 int api_result = XDW_GetPage(h, i, in_path, NULL);
|
|
|
61 if (api_result < 0) {
|
|
|
62 printf("XDW Error: cannot get page\n");
|
|
|
63 return;
|
|
|
64 }
|
|
|
65 }
|
|
|
66 XDW_CloseDocumentHandle(h, NULL);
|
|
|
67 }
|
|
|
68
|
|
|
69 void xdwmerge(const char* list, const char* output) {
|
|
|
70
|
|
|
71 FILE *fp;
|
|
|
72 if ((fp = fopen(list, "r")) == NULL) {
|
|
|
73 fprintf(stderr, "can't open file [%s]\n", list);
|
|
|
74 exit(1);
|
|
|
75 }
|
|
|
76
|
|
|
77 char *all_path = (char*)malloc(MAXLINE * sizeof(char) * _MAX_PATH);
|
|
|
78 if (all_path == NULL) {
|
|
|
79 fprintf(stderr, "can't allocate memory\n");
|
|
|
80 exit(1);
|
|
|
81 }
|
|
|
82
|
|
|
83 int n = 0;
|
|
|
84 char *q;
|
|
|
85 char buf[_MAX_PATH];
|
|
|
86
|
|
|
87 while (fgets(buf, sizeof buf, fp)) {
|
|
|
88 if ((q = strchr(buf, '\n')) != NULL) {
|
|
|
89 *q = '\0';
|
|
|
90 }
|
|
|
91 _fullpath(buf, buf, _MAX_PATH);
|
|
|
92 strncpy(&all_path[n * _MAX_PATH], buf, _MAX_PATH);
|
|
|
93 n++;
|
|
|
94 }
|
|
|
95 fclose(fp);
|
|
|
96
|
|
|
97 char *blk_path = (char*)malloc(BLOCKSZ * sizeof(char) * _MAX_PATH);
|
|
|
98 const char **blk_path_addr = (const char**)malloc((n / BLOCKSZ + 1) * sizeof(char*) * _MAX_PATH);
|
|
|
99 if (blk_path == NULL || blk_path_addr == NULL) {
|
|
|
100 fprintf(stderr, "can't allocate memory\n");
|
|
|
101 exit(1);
|
|
|
102 }
|
|
|
103
|
|
|
104 // process by block
|
|
|
105 int api_result;
|
|
|
106 int bn = 0;
|
|
|
107 for (int p = 0, m = 0; p < n; p++) {
|
|
|
108 m = p % BLOCKSZ;
|
|
|
109 if (m == 0 && p > 0) {
|
|
|
110 sprintf(buf, "fctmp_b%04d.xdw", ++bn);
|
|
|
111 _fullpath(buf, buf, _MAX_PATH);
|
|
|
112 api_result = XDW_MergeXdwFiles(blk_path_addr, BLOCKSZ, buf, NULL);
|
|
|
113 if (api_result < 0) {
|
|
|
114 fprintf(stderr, "can't merge [1] (p = %d, m = %d)\n", p, m);
|
|
|
115 exit(1);
|
|
|
116 }
|
|
|
117 }
|
|
|
118 strncpy(&blk_path[m * _MAX_PATH], &all_path[p * _MAX_PATH], _MAX_PATH);
|
|
|
119 blk_path_addr[m] = &blk_path[m * _MAX_PATH];
|
|
|
120 }
|
|
|
121
|
|
|
122 sprintf(buf, "fctmp_b%04d.xdw", ++bn);
|
|
|
123 _fullpath(buf, buf, _MAX_PATH);
|
|
|
124 int mod = n % BLOCKSZ;
|
|
|
125 if (mod == 0) mod = BLOCKSZ;
|
|
|
126 api_result = XDW_MergeXdwFiles(blk_path_addr, mod, buf, NULL);
|
|
|
127 if (api_result < 0) {
|
|
|
128 fprintf(stderr, "can't merge [2]\n");
|
|
|
129 exit(1);
|
|
|
130 }
|
|
|
131
|
|
|
132 // merge blocks
|
|
|
133 for (int b = 0; b < bn; b++) {
|
|
|
134 sprintf(buf, "fctmp_b%04d.xdw", b + 1);
|
|
|
135 _fullpath(buf, buf, _MAX_PATH);
|
|
|
136 strncpy(&blk_path[b * _MAX_PATH], buf, _MAX_PATH);
|
|
|
137 blk_path_addr[b] = &blk_path[b * _MAX_PATH];
|
|
|
138 }
|
|
|
139 _fullpath(buf, output, _MAX_PATH );
|
|
|
140 api_result = XDW_MergeXdwFiles(blk_path_addr, bn, buf, NULL);
|
|
|
141 if (api_result < 0) {
|
|
|
142 fprintf(stderr, "can't merge [3]\n");
|
|
|
143 exit(1);
|
|
|
144 }
|
|
|
145
|
|
|
146 free(all_path);
|
|
|
147 free(blk_path);
|
|
|
148 free(blk_path_addr);
|
|
|
149 }
|
|
|
150
|
|
|
151 int xdwaddannotation(XDW_DOCUMENT_HANDLE h, int page, int x, int y, char* string, int* sz, int tr)
|
|
|
152 {
|
|
|
153 XDW_ANNOTATION_HANDLE annoation;
|
|
|
154 int api_result = XDW_AddAnnotation(h, XDW_AID_TEXT, page, x, y, NULL, &annoation, NULL);
|
|
|
155 if (api_result < 0) {
|
|
|
156 fprintf(stderr, "can't make annotation\n");
|
|
|
157 return -1;
|
|
|
158 }
|
|
|
159
|
|
|
160 api_result = XDW_SetAnnotationAttribute(h, annoation, XDW_ATN_Text, XDW_ATYPE_STRING, string, 0, NULL);
|
|
|
161 api_result = XDW_SetAnnotationAttribute(h, annoation, XDW_ATN_FontSize, XDW_ATYPE_INT, (char*)(sz), 0, NULL);
|
|
|
162 if (tr) {
|
|
|
163 int color = XDW_COLOR_NONE;
|
|
|
164 api_result = XDW_SetAnnotationAttribute(h, annoation, XDW_ATN_BackColor, XDW_ATYPE_INT, (char*)(&color), 0, NULL);
|
|
|
165 }
|
|
|
166 return 0;
|
|
|
167 }
|
|
|
168
|
|
|
169 void xdwaddpage(const char* file, int sp) {
|
|
|
170 XDW_DOCUMENT_HANDLE h = NULL;
|
|
|
171 XDW_OPEN_MODE_EX mode = {sizeof(XDW_OPEN_MODE_EX), XDW_OPEN_UPDATE, XDW_AUTH_NODIALOGUE};
|
|
|
172 int api_result = XDW_OpenDocumentHandle(file, &h, (XDW_OPEN_MODE*)&mode);
|
|
|
173 if (api_result < 0) {
|
|
|
174 fprintf(stderr, "can't open file\n");
|
|
|
175 exit(1);
|
|
|
176 }
|
|
|
177
|
|
|
178 XDW_DOCUMENT_INFO info = {sizeof(XDW_DOCUMENT_INFO), 0};
|
|
|
179 XDW_GetDocumentInformation(h, &info);
|
|
|
180 int last_page = info.nPages;
|
|
|
181
|
|
|
182 int sz = 80;
|
|
|
183 int tr = 1;
|
|
|
184 char pagenum[12];
|
|
|
185
|
|
|
186 for (int p = 0; p < last_page; p++) {
|
|
|
187 sprintf(pagenum, "%05d-%05d", p + sp, last_page);
|
|
|
188 api_result = xdwaddannotation(h, p + 1, 1769, 5658, pagenum, &sz, tr);
|
|
|
189 api_result = xdwaddannotation(h, p + 1, 18297, 28326, pagenum, &sz, tr);
|
|
|
190 if (api_result < 0) break;
|
|
|
191 }
|
|
|
192
|
|
|
193 if (api_result >= 0) api_result = XDW_SaveDocument(h, NULL);
|
|
|
194
|
|
|
195 XDW_CloseDocumentHandle(h, NULL);
|
|
|
196 }
|
|
|
197
|
|
|
198
|
|
|
199 */
|
|
|
200 import "C"
|
|
|
201
|
|
|
202 import (
|
|
|
203 "encoding/csv"
|
|
|
204 "flag"
|
|
|
205 "fmt"
|
|
|
206 "io"
|
|
|
207 "log"
|
|
|
208 "os"
|
|
|
209 "regexp"
|
|
|
210 "strings"
|
|
|
211 "time"
|
|
|
212
|
|
|
213 "golang.org/x/text/encoding/japanese"
|
|
|
214 "golang.org/x/text/transform"
|
|
|
215 )
|
|
|
216
|
|
|
217 type Data struct {
|
|
|
218 Hno string
|
|
|
219 Page string
|
|
|
220 }
|
|
|
221
|
|
|
222 func (d *Data) ToCsv() string {
|
|
|
223 s := []string{d.Hno, d.Page}
|
|
|
224 return strings.Join(s, ",")
|
|
|
225 }
|
|
|
226
|
|
|
227 var (
|
|
|
228 ver = "0.1"
|
|
|
229
|
|
|
230 clean bool
|
|
|
231 skip bool
|
|
|
232 debug bool
|
|
|
233 xdwfile string
|
|
|
234
|
|
|
235 txtfile = "fctmp_txt.txt"
|
|
|
236 infofile = "fctmp_info.csv"
|
|
|
237 orderfile = "sort.list" // input
|
|
|
238 pagefile = "fctmp_page.txt"
|
|
|
239 output = "output.xdw"
|
|
|
240 )
|
|
|
241
|
|
|
242 func init() {
|
|
|
243 /* INITIALIZE FLAGS */
|
|
|
244 flag.BoolVar(&clean, "c", false, "clean temporary files & exit")
|
|
|
245 flag.BoolVar(&skip, "e", false, "use existed files")
|
|
|
246 flag.BoolVar(&debug, "d", false, "debug mode")
|
|
|
247 flag.StringVar(&xdwfile, "i", "fc.xdw", "target xdw file")
|
|
|
248 }
|
|
|
249
|
|
|
250 func main() {
|
|
|
251 flag.Parse()
|
|
|
252
|
|
|
253 /* PRINT HEADER */
|
|
|
254 fmt.Println("===================================================")
|
|
|
255 fmt.Println(" 普徴のみならず納通なら被保番でソートできるかも... ")
|
|
|
256 fmt.Printf(" - fcgo [ver %s] -\n", ver)
|
|
|
257 fmt.Println("===================================================\n")
|
|
|
258 print_time("now")
|
|
|
259 fmt.Println("[0] start ...")
|
|
|
260
|
|
|
261 /* CLEAN TEMPORARY DIRECTORY */
|
|
|
262 os.Remove(output)
|
|
|
263 if !skip {
|
|
|
264 clean_full()
|
|
|
265 }
|
|
|
266 if clean {
|
|
|
267 os.Exit(0)
|
|
|
268 }
|
|
|
269 print_time("check done")
|
|
|
270
|
|
|
271 fmt.Println("[1] extract ...")
|
|
|
272 if !skip {
|
|
|
273 C.xdw2txt(C.CString(xdwfile), C.CString(txtfile))
|
|
|
274 }
|
|
|
275 print_time("extract done.")
|
|
|
276
|
|
|
277 c := make(chan int)
|
|
|
278 fmt.Println("[2] split ... (run background)")
|
|
|
279 go func() {
|
|
|
280 if !skip {
|
|
|
281 C.xdwsplit1(C.CString(xdwfile))
|
|
|
282 }
|
|
|
283 print_time("split done.")
|
|
|
284 c <- 1
|
|
|
285 }()
|
|
|
286
|
|
|
287 fmt.Println("[3] analize ...")
|
|
|
288 data_hash, err := analize(txtfile, infofile)
|
|
|
289 if err != nil {
|
|
|
290 log.Fatal(err)
|
|
|
291 }
|
|
|
292 print_time("analize done.")
|
|
|
293 debug_print(debug, fmt.Sprintf("len = %d", len(data_hash)))
|
|
|
294
|
|
|
295 fmt.Println("[4] read order ...")
|
|
|
296 h_order, n, err := read_order(orderfile)
|
|
|
297 if err != nil {
|
|
|
298 log.Fatal(err)
|
|
|
299 }
|
|
|
300 print_time("read order done.")
|
|
|
301 debug_print(debug, fmt.Sprintf("len = %d / %d", len(h_order), n))
|
|
|
302
|
|
|
303 fmt.Println("[5] make_list ...")
|
|
|
304 n, err = make_list(h_order, data_hash, pagefile)
|
|
|
305 if err != nil {
|
|
|
306 log.Fatal(err)
|
|
|
307 }
|
|
|
308 print_time("make list done.")
|
|
|
309 debug_print(debug, fmt.Sprintf("order = %d", n))
|
|
|
310
|
|
|
311 <-c
|
|
|
312
|
|
|
313 fmt.Println("[6] merge ...")
|
|
|
314 clean_mini()
|
|
|
315 C.xdwmerge(C.CString(pagefile), C.CString(output))
|
|
|
316 print_time("merge done.")
|
|
|
317 clean_mini()
|
|
|
318
|
|
|
319 fmt.Println("[7] page ...")
|
|
|
320 C.xdwaddpage(C.CString(output), C.int(1))
|
|
|
321 print_time("page done.")
|
|
|
322
|
|
|
323 if !debug {
|
|
|
324 fmt.Println("[8] clean ...")
|
|
|
325 clean_full()
|
|
|
326 print_time("clean done.")
|
|
|
327 }
|
|
|
328 }
|
|
|
329
|
|
|
330 func analize(txtfile, infofile string) (map[string]Data, error) {
|
|
|
331 hash := make(map[string]Data)
|
|
|
332
|
|
|
333 c, err := os.ReadFile(txtfile)
|
|
|
334 if err != nil {
|
|
|
335 return hash, err
|
|
|
336 }
|
|
|
337 r := strings.NewReader(string(c))
|
|
|
338 tr := transform.NewReader(r, japanese.ShiftJIS.NewDecoder())
|
|
|
339 b, err := io.ReadAll(tr)
|
|
|
340 if err != nil {
|
|
|
341 return hash, err
|
|
|
342 }
|
|
|
343
|
|
|
344 regHno := regexp.MustCompile(`0[1238]0[0-9]{7}`)
|
|
|
345
|
|
|
346 var csv string
|
|
|
347
|
|
|
348 buf := strings.Split(string(b), "期別保険料額")
|
|
|
349 for p, v := range buf[:len(buf)-1] {
|
|
|
350 s := zen2han(v)
|
|
|
351 hno := regHno.FindString(s)
|
|
|
352 page := fmt.Sprintf("%05d", p + 1)
|
|
|
353
|
|
|
354 data := Data {
|
|
|
355 Hno: hno,
|
|
|
356 Page: page,
|
|
|
357 }
|
|
|
358 hash[hno] = data
|
|
|
359
|
|
|
360 csv += data.ToCsv() + "\n"
|
|
|
361 }
|
|
|
362
|
|
|
363 if err := os.WriteFile(infofile, []byte(csv), 0644); err != nil {
|
|
|
364 return hash, err
|
|
|
365 }
|
|
|
366 return hash, nil;
|
|
|
367 }
|
|
|
368
|
|
|
369 func zen2han(s string) (string) {
|
|
|
370 s = strings.ReplaceAll(s, "０", "0")
|
|
|
371 s = strings.ReplaceAll(s, "１", "1")
|
|
|
372 s = strings.ReplaceAll(s, "２", "2")
|
|
|
373 s = strings.ReplaceAll(s, "３", "3")
|
|
|
374 s = strings.ReplaceAll(s, "４", "4")
|
|
|
375 s = strings.ReplaceAll(s, "５", "5")
|
|
|
376 s = strings.ReplaceAll(s, "６", "6")
|
|
|
377 s = strings.ReplaceAll(s, "７", "7")
|
|
|
378 s = strings.ReplaceAll(s, "８", "8")
|
|
|
379 return strings.ReplaceAll(s, "９", "9")
|
|
|
380 }
|
|
|
381
|
|
|
382 func read_order(csvfile string) ([]string, int, error) {
|
|
|
383 var order []string
|
|
|
384
|
|
|
385 c, err := os.ReadFile(csvfile)
|
|
|
386 if err != nil {
|
|
|
387 return order, -1, err
|
|
|
388 }
|
|
|
389 r := strings.NewReader(string(c))
|
|
|
390 tr := transform.NewReader(r, japanese.ShiftJIS.NewDecoder())
|
|
|
391 b, err := io.ReadAll(tr)
|
|
|
392 if err != nil {
|
|
|
393 return order, -1, err
|
|
|
394 }
|
|
|
395
|
|
|
396 n := 0
|
|
|
397 cr := csv.NewReader(strings.NewReader(string(b)))
|
|
|
398 for {
|
|
|
399 record, err := cr.Read()
|
|
|
400 if err == io.EOF {
|
|
|
401 break
|
|
|
402 }
|
|
|
403 if err != nil {
|
|
|
404 return order, n, err
|
|
|
405 }
|
|
|
406
|
|
|
407 s := zen2han(record[0])
|
|
|
408 if strings.HasPrefix(s, "0") {
|
|
|
409 order = append(order, s)
|
|
|
410 }
|
|
|
411 n++
|
|
|
412 }
|
|
|
413 return order, n, nil
|
|
|
414 }
|
|
|
415
|
|
|
416 func make_list(h_order []string, hash map[string]Data, pagefile string) (int, error) {
|
|
|
417 var n int
|
|
|
418 var list []string
|
|
|
419 done := make(map[string]bool)
|
|
|
420
|
|
|
421 for _, h := range h_order {
|
|
|
422 if _, ok := done[h]; !ok {
|
|
|
423 if data, ok := hash[h]; ok {
|
|
|
424 list = append(list, data.Page)
|
|
|
425 done[h] = true
|
|
|
426 n++
|
|
|
427 }
|
|
|
428 }
|
|
|
429 }
|
|
|
430
|
|
|
431 if err := write_pagefile(pagefile, list); err != nil {
|
|
|
432 return n, err
|
|
|
433 }
|
|
|
434 return n, nil
|
|
|
435 }
|
|
|
436
|
|
|
437 func write_pagefile(file string, list []string) error {
|
|
|
438 f, err := os.OpenFile(file, os.O_RDWR|os.O_CREATE, 0755)
|
|
|
439 if err != nil {
|
|
|
440 return err
|
|
|
441 }
|
|
|
442 for _, p := range list {
|
|
|
443 fmt.Fprintf(f, "fctmp_%s.xdw\n", p)
|
|
|
444 }
|
|
|
445 if err := f.Close(); err != nil {
|
|
|
446 return err
|
|
|
447 }
|
|
|
448 return nil
|
|
|
449 }
|
|
|
450
|
|
|
451 func print_time(msg string) {
|
|
|
452 now := time.Now()
|
|
|
453 fmt.Printf("\t%v # %s\n", now, msg)
|
|
|
454 }
|
|
|
455
|
|
|
456 func debug_print(debug bool, msg string) {
|
|
|
457 if debug {
|
|
|
458 fmt.Printf("\t%s\n", msg)
|
|
|
459 }
|
|
|
460 }
|
|
|
461
|
|
|
462 func clean_full() error {
|
|
|
463 return clean_file("fctmp_")
|
|
|
464 }
|
|
|
465
|
|
|
466 func clean_mini() error {
|
|
|
467 return clean_file("fctmp_b")
|
|
|
468 }
|
|
|
469
|
|
|
470 func clean_file(prefix string) error {
|
|
|
471 files, err := os.ReadDir(".")
|
|
|
472 if err != nil {
|
|
|
473 return err
|
|
|
474 }
|
|
|
475
|
|
|
476 for _, file := range files {
|
|
|
477 if strings.HasPrefix(file.Name(), prefix) {
|
|
|
478 os.Remove(file.Name())
|
|
|
479 continue
|
|
|
480 }
|
|
|
481 }
|
|
|
482 return err
|
|
|
483 }
|
|
|
484
|
