src/stf-parse.h

   1 /* vim: set sw=8: -*- Mode: C; tab-width: 8; indent-tabs-mode: t; c-basic-offset: 8 -*- */
   2 #ifndef _GNM_STF_PARSE_H_
   3 # define _GNM_STF_PARSE_H_
   4
   5 #include <glib.h>
   6 #include <gnumeric.h>
   7
   8 G_BEGIN_DECLS
   9
  10 typedef enum {
  11         PARSE_TYPE_NOTSET    = 1 << 0,
  12         PARSE_TYPE_CSV       = 1 << 1,
  13         PARSE_TYPE_FIXED     = 1 << 2
  14 } StfParseType_t;
  15
  16 /* Additive.  */
  17 typedef enum {
  18         TRIM_TYPE_NEVER      = 0,
  19         TRIM_TYPE_LEFT       = 1 << 0,
  20         TRIM_TYPE_RIGHT      = 1 << 1
  21 } StfTrimType_t;
  22
  23 typedef struct {
  24         StfParseType_t       parsetype;             /* The type of import to do */
  25         StfTrimType_t        trim_spaces;           /* Trim spaces in fields? */
  26
  27         GSList *             terminator;            /* Line terminators */
  28         char *               locale;
  29
  30         struct _StfCompiledTerminator {
  31                 guchar       min, max;
  32         } compiled_terminator;
  33
  34         /* CSV related */
  35         struct _StfSeparator {
  36                 GSList *str;
  37                 char   *chr;
  38                 gboolean duplicates;         /* See two text separators as one? */
  39         } sep;
  40         gunichar             stringindicator;       /* String indicator */
  41         gboolean             indicator_2x_is_single;/* 2 quote chars form a single non-terminating quote */
  42         gboolean             trim_seps;             /* Ignore initial seps.  */
  43
  44         /* Fixed width related */
  45         GArray              *splitpositions;        /* Positions where text will be split vertically */
  46
  47         gboolean             *col_autofit_array;    /* 0/1 array indicating  */
  48                                                     /* which col widths to autofit  */
  49         gboolean             *col_import_array;     /* 0/1 array indicating  */
  50                                                     /* which cols to import  */
  51         unsigned int         col_import_array_len;
  52
  53         GPtrArray            *formats;              /* Contains GOFormat *s */
  54         GPtrArray            *formats_decimal;      /* Contains GString *s */
  55         GPtrArray            *formats_thousand;     /* Contains GString *s */
  56         GPtrArray            *formats_curr;         /* Contains GString *s */
  57
  58         gboolean             cols_exceeded;         /* This is set to TRUE if */
  59                                                     /* we tried to import more than */
  60                                                     /* SHEET_MAX_COLS columns */
  61         gboolean             rows_exceeded;         /* Ditto rows.  */
  62         unsigned             ref_count;             /* Boxed type */
  63 } StfParseOptions_t;
  64
  65 /* CREATION/DESTRUCTION of stf options struct */
  66
  67 GType               stf_parse_options_get_type                        (void);
  68 void                stf_parse_options_free                            (StfParseOptions_t *parseoptions);
  69
  70 StfParseOptions_t  *stf_parse_options_guess                           (char const *data);
  71 StfParseOptions_t  *stf_parse_options_guess_csv                       (char const *data);
  72 void                stf_parse_options_guess_formats                   (StfParseOptions_t *po,
  73                                                                        char const *data);
  74
  75 /* MANIPULATION of stf options struct */
  76
  77 void stf_parse_options_set_type                        (StfParseOptions_t *parseoptions,
  78                                                         StfParseType_t const parsetype);
  79 void stf_parse_options_clear_line_terminator           (StfParseOptions_t *parseoptions);
  80 void stf_parse_options_add_line_terminator             (StfParseOptions_t *parseoptions,
  81                                                         char const *terminator);
  82 void stf_parse_options_set_trim_spaces                 (StfParseOptions_t *parseoptions,
  83                                                         StfTrimType_t const trim_spaces);
  84 void stf_parse_options_csv_set_separators              (StfParseOptions_t *parseoptions,
  85                                                         char const *character, GSList const *seps);
  86 void stf_parse_options_csv_set_stringindicator         (StfParseOptions_t *parseoptions,
  87                                                         gunichar const stringindicator);
  88 void stf_parse_options_csv_set_indicator_2x_is_single  (StfParseOptions_t *parseoptions,
  89                                                         gboolean const indic_2x);
  90 void stf_parse_options_csv_set_duplicates              (StfParseOptions_t *parseoptions,
  91                                                         gboolean const duplicates);
  92 void stf_parse_options_csv_set_trim_seps               (StfParseOptions_t *parseoptions,
  93                                                         gboolean const trim_seps);
  94 void stf_parse_options_fixed_splitpositions_clear      (StfParseOptions_t *parseoptions);
  95 void stf_parse_options_fixed_splitpositions_add        (StfParseOptions_t *parseoptions,
  96                                                         int position);
  97 void stf_parse_options_fixed_splitpositions_remove     (StfParseOptions_t *parseoptions,
  98                                                         int position);
  99 int stf_parse_options_fixed_splitpositions_count       (StfParseOptions_t *parseoptions);
 100 int stf_parse_options_fixed_splitpositions_nth         (StfParseOptions_t *parseoptions, int n);
 101
 102 /* USING the stf structs to actually do some parsing, these are the lower-level functions and utility functions */
 103
 104 GPtrArray       *stf_parse_general                      (StfParseOptions_t *parseoptions,
 105                                                          GStringChunk *lines_chunk,
 106                                                          char const *data,
 107                                                          char const *data_end);
 108 void             stf_parse_general_free                 (GPtrArray *lines);
 109 GPtrArray       *stf_parse_lines                        (StfParseOptions_t *parseoptions,
 110                                                          GStringChunk *lines_chunk,
 111                                                          char const *data,
 112                                                          int maxlines,
 113                                                          gboolean with_lineno);
 114
 115 void             stf_parse_options_fixed_autodiscover   (StfParseOptions_t *parseoptions,
 116                                                          char const *data,
 117                                                          char const *data_end);
 118
 119 char const      *stf_parse_find_line                    (StfParseOptions_t *parseoptions,
 120                                                          char const *data,
 121                                                          int line);
 122
 123 /* Higher level functions, can be used for directly parsing into an application specific data container */
 124 gboolean         stf_parse_sheet                        (StfParseOptions_t *parseoptions,
 125                                                          char const *data, char const *data_end,
 126                                                          Sheet *sheet,
 127                                                          int start_col, int start_row);
 128
 129 GnmCellRegion   *stf_parse_region                       (StfParseOptions_t *parseoptions,
 130                                                          char const *data, char const *data_end,
 131                                                          Workbook const *wb);
 132
 133 G_END_DECLS
 134
 135 #endif /* _GNM_STF_PARSE_H_ */