Search

A file you can't search is a file you can only scroll. The good news after the last few lessons: search is easy — plain C string work, no escape codes, no state machines. The design still deserves five minutes, because search UX has sharp conventions.

Find-next semantics

The operation that matters is not "find" but find next: given where I am, where is the following match? Repeat-invocations then walk match to match. Two conventions define it:

• Start searching at a given position (inclusive), and let the caller pass "cursor + one column" for repeat-search — else pressing find-next while standing on a match finds the same match forever. Keeping the +1 out of the search function keeps the function honest and the policy visible at the call site.
• Wrap around. Hitting the last match then searching again jumps back to the first (vim flashes "search hit BOTTOM, continuing at TOP"). Implement by scanning position → end of buffer, then start of buffer → position.

Within a line, strstr(3) does the byte work — find the first occurrence of a needle in a haystack. The only wrinkle: starting mid-line means searching the line's suffix (strstr(line + col, q)), then reporting the match's column relative to the whole line (add the offset back — a classic pointer-arithmetic fencepost).

Real editors layer niceties on this core — case folding, highlight-all, incremental search that jumps as you type, regex. All of them sit on exactly this function; incremental search, for instance, is just re-running find-next from the search's origin on every keystroke of the query.

int editor_find_next(const struct editor *e, const char *query,
                     int from_row, int from_col,
                     int *match_row, int *match_col);