prism 2.0.0-arm64-darwin

data/src/string_query.c

@@ -0,0 +1,180 @@
+#include "prism/string_query.h"
+
+#include "prism/internal/char.h"
+#include "prism/internal/encoding.h"
+#include "prism/internal/parser.h"
+
+#include <assert.h>
+#include <string.h>
+
+/** The category of slice returned from pm_slice_type. */
+typedef enum {
+    /** Returned when the given encoding name is invalid. */
+    PM_SLICE_TYPE_ERROR = -1,
+
+    /** Returned when no other types apply to the slice. */
+    PM_SLICE_TYPE_NONE,
+
+    /** Returned when the slice is a valid local variable name. */
+    PM_SLICE_TYPE_LOCAL,
+
+    /** Returned when the slice is a valid constant name. */
+    PM_SLICE_TYPE_CONSTANT,
+
+    /** Returned when the slice is a valid method name. */
+    PM_SLICE_TYPE_METHOD_NAME
+} pm_slice_type_t;
+
+/**
+ * Check that the slice is a valid local variable name or constant.
+ */
+static pm_slice_type_t
+pm_slice_type(const uint8_t *source, size_t length, const char *encoding_name) {
+    // first, get the right encoding object
+    const pm_encoding_t *encoding = pm_encoding_find((const uint8_t *) encoding_name, (const uint8_t *) (encoding_name + strlen(encoding_name)));
+    if (encoding == NULL) return PM_SLICE_TYPE_ERROR;
+
+    // check that there is at least one character
+    if (length == 0) return PM_SLICE_TYPE_NONE;
+
+    size_t width;
+    if ((width = encoding->alpha_char(source, (ptrdiff_t) length)) != 0) {
+        // valid because alphabetical
+    } else if (*source == '_') {
+        // valid because underscore
+        width = 1;
+    } else if ((*source >= 0x80) && ((width = encoding->char_width(source, (ptrdiff_t) length)) > 0)) {
+        // valid because multibyte
+    } else {
+        // invalid because no match
+        return PM_SLICE_TYPE_NONE;
+    }
+
+    // determine the type of the slice based on the first character
+    const uint8_t *end = source + length;
+    pm_slice_type_t result = encoding->isupper_char(source, end - source) ? PM_SLICE_TYPE_CONSTANT : PM_SLICE_TYPE_LOCAL;
+
+    // next, iterate through all of the bytes of the string to ensure that they
+    // are all valid identifier characters
+    source += width;
+
+    while (source < end) {
+        if ((width = encoding->alnum_char(source, end - source)) != 0) {
+            // valid because alphanumeric
+            source += width;
+        } else if (*source == '_') {
+            // valid because underscore
+            source++;
+        } else if ((*source >= 0x80) && ((width = encoding->char_width(source, end - source)) > 0)) {
+            // valid because multibyte
+            source += width;
+        } else {
+            // invalid because no match
+            break;
+        }
+    }
+
+    // accept a ! or ? at the end of the slice as a method name
+    if (*source == '!' || *source == '?' || *source == '=') {
+        source++;
+        result = PM_SLICE_TYPE_METHOD_NAME;
+    }
+
+    // valid if we are at the end of the slice
+    return source == end ? result : PM_SLICE_TYPE_NONE;
+}
+
+/**
+ * Check that the slice is a valid local variable name.
+ */
+pm_string_query_t
+pm_string_query_local(const uint8_t *source, size_t length, const char *encoding_name) {
+    switch (pm_slice_type(source, length, encoding_name)) {
+        case PM_SLICE_TYPE_ERROR:
+            return PM_STRING_QUERY_ERROR;
+        case PM_SLICE_TYPE_NONE:
+        case PM_SLICE_TYPE_CONSTANT:
+        case PM_SLICE_TYPE_METHOD_NAME:
+            return PM_STRING_QUERY_FALSE;
+        case PM_SLICE_TYPE_LOCAL:
+            return PM_STRING_QUERY_TRUE;
+    }
+
+    assert(false && "unreachable");
+    return PM_STRING_QUERY_FALSE;
+}
+
+/**
+ * Check that the slice is a valid constant name.
+ */
+pm_string_query_t
+pm_string_query_constant(const uint8_t *source, size_t length, const char *encoding_name) {
+    switch (pm_slice_type(source, length, encoding_name)) {
+        case PM_SLICE_TYPE_ERROR:
+            return PM_STRING_QUERY_ERROR;
+        case PM_SLICE_TYPE_NONE:
+        case PM_SLICE_TYPE_LOCAL:
+        case PM_SLICE_TYPE_METHOD_NAME:
+            return PM_STRING_QUERY_FALSE;
+        case PM_SLICE_TYPE_CONSTANT:
+            return PM_STRING_QUERY_TRUE;
+    }
+
+    assert(false && "unreachable");
+    return PM_STRING_QUERY_FALSE;
+}
+
+/**
+ * Check that the slice is a valid method name.
+ */
+pm_string_query_t
+pm_string_query_method_name(const uint8_t *source, size_t length, const char *encoding_name) {
+#define B(p) ((p) ? PM_STRING_QUERY_TRUE : PM_STRING_QUERY_FALSE)
+#define C1(c) (*source == c)
+#define C2(s) (memcmp(source, s, 2) == 0)
+#define C3(s) (memcmp(source, s, 3) == 0)
+
+    switch (pm_slice_type(source, length, encoding_name)) {
+        case PM_SLICE_TYPE_ERROR:
+            return PM_STRING_QUERY_ERROR;
+        case PM_SLICE_TYPE_NONE:
+            break;
+        case PM_SLICE_TYPE_LOCAL:
+            // numbered parameters are not valid method names
+            return B((length != 2) || (source[0] != '_') || (source[1] == '0') || !pm_char_is_decimal_digit(source[1]));
+        case PM_SLICE_TYPE_CONSTANT:
+            // all constants are valid method names
+        case PM_SLICE_TYPE_METHOD_NAME:
+            // all method names are valid method names
+            return PM_STRING_QUERY_TRUE;
+    }
+
+    switch (length) {
+        case 1:
+            return B(C1('&') || C1('`') || C1('!') || C1('^') || C1('>') || C1('<') || C1('-') || C1('%') || C1('|') || C1('+') || C1('/') || C1('*') || C1('~'));
+        case 2:
+            return B(C2("!=") || C2("!~") || C2("[]") || C2("==") || C2("=~") || C2(">=") || C2(">>") || C2("<=") || C2("<<") || C2("**"));
+        case 3:
+            return B(C3("===") || C3("<=>") || C3("[]="));
+        default:
+            return PM_STRING_QUERY_FALSE;
+    }
+
+#undef B
+#undef C1
+#undef C2
+#undef C3
+}
+
+/**
+ * Check that the slice is a keyword. Keywords are spelled in ASCII, so any
+ * ASCII-compatible encoding spells them with the same bytes.
+ */
+pm_string_query_t
+pm_string_query_keyword(const uint8_t *source, size_t length, const char *encoding_name) {
+    if (pm_encoding_find((const uint8_t *) encoding_name, (const uint8_t *) (encoding_name + strlen(encoding_name))) == NULL) {
+        return PM_STRING_QUERY_ERROR;
+    }
+
+    return pm_keyword_p(source, source + length) ? PM_STRING_QUERY_TRUE : PM_STRING_QUERY_FALSE;
+}

data/src/stringy.c

@@ -0,0 +1,61 @@
+#include "prism/internal/stringy.h"
+
+#include <assert.h>
+#include <string.h>
+
+/**
+ * Initialize a shared string that is based on initial input.
+ */
+void
+pm_string_shared_init(pm_string_t *string, const uint8_t *start, const uint8_t *end) {
+    assert(start <= end);
+
+    *string = (pm_string_t) {
+        .source = start,
+        .length = (size_t) (end - start)
+    };
+}
+
+/**
+ * Initialize a string that views the given bytes.
+ */
+void
+pm_string_init(pm_string_t *string, const uint8_t *source, size_t length) {
+    *string = (pm_string_t) { .source = source, .length = length };
+}
+
+/**
+ * Compare the underlying lengths and bytes of two strings. Returns 0 if the
+ * strings are equal, a negative number if the left string is less than the
+ * right string, and a positive number if the left string is greater than the
+ * right string.
+ */
+int
+pm_string_compare(const pm_string_t *left, const pm_string_t *right) {
+    size_t left_length = pm_string_length(left);
+    size_t right_length = pm_string_length(right);
+
+    if (left_length < right_length) {
+        return -1;
+    } else if (left_length > right_length) {
+        return 1;
+    }
+
+    return memcmp(pm_string_source(left), pm_string_source(right), left_length);
+}
+
+/**
+ * Returns the length associated with the string.
+ */
+size_t
+pm_string_length(const pm_string_t *string) {
+    return string->length;
+}
+
+/**
+ * Returns the start pointer associated with the string.
+ */
+const uint8_t *
+pm_string_source(const pm_string_t *string) {
+    return string->source;
+}

data/src/strncasecmp.c

@@ -0,0 +1,37 @@
+#include "prism/internal/strncasecmp.h"
+
+#include "prism/compiler/inline.h"
+
+/**
+ * A locale-insensitive version of `tolower(3)`
+ */
+static PRISM_INLINE int
+pm_tolower(int c) {
+    if ('A' <= c && c <= 'Z') {
+        return c | 0x20;
+    }
+    return c;
+}
+
+/**
+ * Compare two strings, ignoring case, up to the given length. Returns 0 if the
+ * strings are equal, a negative number if string1 is less than string2, or a
+ * positive number if string1 is greater than string2.
+ *
+ * Note that this is effectively our own implementation of strncasecmp, but it's
+ * not available on all of the platforms we want to support so we're rolling it
+ * here.
+ */
+int
+pm_strncasecmp(const uint8_t *string1, const uint8_t *string2, size_t length) {
+    size_t offset = 0;
+    int difference = 0;
+
+    while (offset < length && string1[offset] != '\0') {
+        if (string2[offset] == '\0') return string1[offset];
+        if ((difference = pm_tolower(string1[offset]) - pm_tolower(string2[offset])) != 0) return difference;
+        offset++;
+    }
+
+    return difference;
+}