1 files changed, 143 insertions, 0 deletions
diff --git a/src/spreed.c b/src/spreed.c
new file mode 100755
index 0000000..933bec7
--- /dev/null
+++ b/src/spreed.c
@@ -0,0 +1,143 @@
+#if 0
+${CC:-gcc} -s -O2 -std=c99 -Wall -o ${SPREED_INSTALL_DIR:-.}/spreed src/spreed.c
+exit
+#endif
+
+/* TODO
+ * contractions don't work
+ * - result of trying to separate a+b
+ */
+
+#include <locale.h>
+#include <stdint.h>
+#include <stdio.h>
+#include <string.h>
+#include <wchar.h>
+#include <wctype.h>
+
+#define CLR_FG     L"\033[31m"
+#define CLR_RESET  L"\033[0m"
+
+uint8_t get_optimal_recognition_point(size_t len) {
+	static uint8_t orps[9] = {0,0,0,1,1,2,2,2,2};
+	if (len >= 10) return 3;
+	return orps[len-1];
+}
+
+int main(int argc, char *argv[])
+{
+	wchar_t word[48];
+	wchar_t wc;
+	size_t res;
+	int c;
+	char chararr[4] = {0};
+	uint8_t u, uu;
+	size_t ichar = 0, iword = 0;
+	uint8_t b_word_break = 0;
+
+	memset(word, 0, 48 * sizeof(wchar_t));
+	memset(chararr, 0, sizeof(chararr));
+
+	setlocale(LC_ALL, "");
+	fwide(stdout, 1);
+
+	fwprintf(stdout, L"          V\n");
+	do {
+		b_word_break = 0;
+		c = getchar();
+		if (c == EOF) {
+			b_word_break = iword > 0;
+			goto END_OF_WORD;
+		}
+		chararr[ichar++] = (char)c;
+
+		res = mbrtowc(&wc, chararr, 4, NULL);
+		switch(res) {
+			case 0:
+				// null char
+				if (iword == 0 && ichar == 1) {
+					ichar = 0;
+					memset(chararr, 0, 4 * sizeof(char));
+					continue;
+				} else {
+					// fwprintf(stderr, L"cowabunga: %X %X, %lX\n", ichar, c, wc);
+					b_word_break = 1;
+					goto END_OF_WORD;
+				}
+			case (size_t)-2:
+				// incomplete wchar_t, keep reading bytes
+				continue;
+			case (size_t)-1:
+				if (ichar == 4) {
+					// emojis return (size_t)-1 until we have all the bytes
+					fwprintf(stderr, L"\nmbrtowc encoding error\n");
+					return 1;
+				}
+				continue;
+		}
+
+		ichar = 0;
+		memset(chararr, 0, 4 * sizeof(char));
+
+		// not a printable character, swallow
+		if (!iswprint(wc) && iword == 0) {
+			continue;
+		}
+
+		if (iswspace(wc)) {
+			if (iword == 0) {
+				continue;
+			}
+			b_word_break = 1;
+		}
+
+		if (iword > 0) {
+			if (iswpunct(word[iword-1]) != iswpunct(wc)) {
+				b_word_break = 1;
+			}
+		}
+
+END_OF_WORD:
+		ichar = 0;
+
+		if (b_word_break != 0 || iword == 37) {
+			if (iword == 0) {
+				fwprintf(stderr, L"\n0 length word\n");
+				return 2;
+			}
+			ichar = get_optimal_recognition_point(iword);
+			uu = 0;
+			putwchar('\r');
+			for (u = 0; u < 48; ++u) {
+				if (u < 10-ichar) {
+					putwchar(L' ');
+					continue;
+				}
+				if (uu < iword) {
+					if (u == 10) {
+						fwprintf(stdout, L"%ls", CLR_FG);
+					}
+					putwchar(word[uu++]);
+					if (u == 10) {
+						fwprintf(stdout, L"%ls", CLR_RESET);
+					}
+					continue;
+				}
+				putwchar(L' ');
+			}
+			putwchar('\n');
+
+			iword = 0;
+			memset(word, 0, 48 * sizeof(wchar_t));
+		}
+
+		ichar = 0;
+		if (!iswspace(wc)) {
+			word[iword++] = wc;
+		}
+
+	} while (c != EOF);
+	putwchar('\n');
+
+	return 0;
+}