Import ruby/prism@f41ba6d

Author: andrykonchin

src/main/c/yarp/include/prism.h

@@ -21,6 +21,7 @@
 #include "prism/parser.h"
 #include "prism/prettyprint.h"
 #include "prism/regexp.h"
+#include "prism/static_literals.h"
 #include "prism/version.h"
 
 #include <assert.h>

src/main/c/yarp/include/prism/ast.h

@@ -5,6 +5,7 @@
 /* if you are looking to modify the                                           */
 /* template                                                                   */
 /******************************************************************************/
+
 /**
  * @file ast.h
  *

src/main/c/yarp/include/prism/diagnostic.h

@@ -306,7 +306,10 @@ typedef enum {
     PM_WARN_AMBIGUOUS_SLASH,
     PM_WARN_EQUAL_IN_CONDITIONAL,
     PM_WARN_END_IN_METHOD,
+    PM_WARN_DUPLICATED_HASH_KEY,
+    PM_WARN_DUPLICATED_WHEN_CLAUSE,
     PM_WARN_FLOAT_OUT_OF_RANGE,
+    PM_WARN_INTEGER_IN_FLIP_FLOP,
 
     // This is the number of diagnostic codes.
     PM_DIAGNOSTIC_ID_LEN,

src/main/c/yarp/include/prism/node.h

@@ -10,6 +10,17 @@
 #include "prism/parser.h"
 #include "prism/util/pm_buffer.h"
 
+/**
+ * Attempts to grow the node list to the next size. If there is already
+ * capacity in the list, this function does nothing. Otherwise it reallocates
+ * the list to be twice as large as it was before. If the reallocation fails,
+ * this function returns false, otherwise it returns true.
+ *
+ * @param list The list to grow.
+ * @return True if the list was successfully grown, false otherwise.
+ */
+bool pm_node_list_grow(pm_node_list_t *list);
+
 /**
  * Append a new node onto the end of the node list.
  *
@@ -18,6 +29,22 @@
  */
 void pm_node_list_append(pm_node_list_t *list, pm_node_t *node);
 
+/**
+ * Prepend a new node onto the beginning of the node list.
+ *
+ * @param list The list to prepend to.
+ * @param node The node to prepend.
+ */
+void
+pm_node_list_prepend(pm_node_list_t *list, pm_node_t *node);
+
+/**
+ * Free the internal memory associated with the given node list.
+ *
+ * @param list The list to free.
+ */
+void pm_node_list_free(pm_node_list_t *list);
+
 /**
  * Deallocate a node and all of its children.
  *

src/main/c/yarp/include/prism/options.h

@@ -76,10 +76,50 @@ typedef struct {
      */
     pm_options_version_t version;
 
+    /** A bitset of the various options that were set on the command line. */
+    uint8_t command_line;
+
     /** Whether or not the frozen string literal option has been set. */
     bool frozen_string_literal;
 } pm_options_t;
 
+/**
+ * A bit representing whether or not the command line -a option was set. -a
+ * splits the input line $_ into $F.
+ */
+static const uint8_t PM_OPTIONS_COMMAND_LINE_A = 0x1;
+
+/**
+ * A bit representing whether or not the command line -e option was set. -e
+ * allow the user to specify a script to be executed. This is necessary for
+ * prism to know because certain warnings are not generated when -e is used.
+ */
+static const uint8_t PM_OPTIONS_COMMAND_LINE_E = 0x2;
+
+/**
+ * A bit representing whether or not the command line -l option was set. -l
+ * chomps the input line by default.
+ */
+static const uint8_t PM_OPTIONS_COMMAND_LINE_L = 0x4;
+
+/**
+ * A bit representing whether or not the command line -n option was set. -n
+ * wraps the script in a while gets loop.
+ */
+static const uint8_t PM_OPTIONS_COMMAND_LINE_N = 0x8;
+
+/**
+ * A bit representing whether or not the command line -p option was set. -p
+ * prints the value of $_ at the end of each loop.
+ */
+static const uint8_t PM_OPTIONS_COMMAND_LINE_P = 0x10;
+
+/**
+ * A bit representing whether or not the command line -x option was set. -x
+ * searches the input file for a shebang that matches the current Ruby engine.
+ */
+static const uint8_t PM_OPTIONS_COMMAND_LINE_X = 0x20;
+
 /**
  * Set the filepath option on the given options struct.
  *
@@ -112,6 +152,14 @@ PRISM_EXPORTED_FUNCTION void pm_options_encoding_set(pm_options_t *options, cons
  */
 PRISM_EXPORTED_FUNCTION void pm_options_frozen_string_literal_set(pm_options_t *options, bool frozen_string_literal);
 
+/**
+ * Sets the command line option on the given options struct.
+ *
+ * @param options The options struct to set the command line option on.
+ * @param command_line The command_line value to set.
+ */
+PRISM_EXPORTED_FUNCTION void pm_options_command_line_set(pm_options_t *options, uint8_t command_line);
+
 /**
  * Set the version option on the given options struct by parsing the given
  * string. If the string contains an invalid option, this returns false.
@@ -186,7 +234,10 @@ PRISM_EXPORTED_FUNCTION void pm_options_free(pm_options_t *options);
  * | `4`     | the length the encoding    |
  * | ...     | the encoding bytes         |
  * | `1`     | frozen string literal      |
- * | `1`     | suppress warnings          |
+ * | `1`     | -p command line option     |
+ * | `1`     | -n command line option     |
+ * | `1`     | -l command line option     |
+ * | `1`     | -a command line option     |
  * | `1`     | the version                |
  * | `4`     | the number of scopes       |
  * | ...     | the scopes                 |

src/main/c/yarp/include/prism/parser.h

@@ -703,6 +703,9 @@ struct pm_parser {
     /** The version of prism that we should use to parse. */
     pm_options_version_t version;
 
+    /** The command line flags given from the options. */
+    uint8_t command_line;
+
     /** Whether or not we're at the beginning of a command. */
     bool command_start;
 

src/main/c/yarp/include/prism/static_literals.h

@@ -0,0 +1,109 @@
+/**
+ * @file static_literals.h
+ *
+ * A set of static literal nodes that can be checked for duplicates.
+ */
+#ifndef PRISM_STATIC_LITERALS_H
+#define PRISM_STATIC_LITERALS_H
+
+#include "prism/defines.h"
+#include "prism/ast.h"
+#include "prism/node.h"
+#include "prism/parser.h"
+
+#include <assert.h>
+#include <stdbool.h>
+
+/**
+ * An internal hash table for a set of nodes.
+ */
+typedef struct {
+    /** The array of nodes in the hash table. */
+    pm_node_t **nodes;
+
+    /** The size of the hash table. */
+    uint32_t size;
+
+    /** The space that has been allocated in the hash table. */
+    uint32_t capacity;
+} pm_node_hash_t;
+
+/**
+ * Certain sets of nodes (hash keys and when clauses) check for duplicate nodes
+ * to alert the user of potential issues. To do this, we keep a set of the nodes
+ * that have been seen so far, and compare whenever we find a new node.
+ *
+ * We bucket the nodes based on their type to minimize the number of comparisons
+ * that need to be performed.
+ */
+typedef struct {
+    /**
+     * This is the set of IntegerNode and SourceLineNode instances.
+     */
+    pm_node_hash_t integer_nodes;
+
+    /**
+     * This is the set of FloatNode instances.
+     */
+    pm_node_hash_t float_nodes;
+
+    /**
+     * This is the set of RationalNode and ImaginaryNode instances.
+     */
+    pm_node_hash_t number_nodes;
+
+    /**
+     * This is the set of StringNode and SourceFileNode instances.
+     */
+    pm_node_hash_t string_nodes;
+
+    /**
+     * This is the set of RegularExpressionNode instances.
+     */
+    pm_node_hash_t regexp_nodes;
+
+    /**
+     * This is the set of SymbolNode instances.
+     */
+    pm_node_hash_t symbol_nodes;
+
+    /**
+     * A pointer to the last TrueNode instance that was inserted, or NULL.
+     */
+    pm_node_t *true_node;
+
+    /**
+     * A pointer to the last FalseNode instance that was inserted, or NULL.
+     */
+    pm_node_t *false_node;
+
+    /**
+     * A pointer to the last NilNode instance that was inserted, or NULL.
+     */
+    pm_node_t *nil_node;
+
+    /**
+     * A pointer to the last SourceEncodingNode instance that was inserted, or
+     * NULL.
+     */
+    pm_node_t *source_encoding_node;
+} pm_static_literals_t;
+
+/**
+ * Add a node to the set of static literals.
+ *
+ * @param parser The parser that created the node.
+ * @param literals The set of static literals to add the node to.
+ * @param node The node to add to the set.
+ * @return A pointer to the node that is being overwritten, if there is one.
+ */
+pm_node_t * pm_static_literals_add(const pm_parser_t *parser, pm_static_literals_t *literals, pm_node_t *node);
+
+/**
+ * Free the internal memory associated with the given static literals set.
+ *
+ * @param literals The set of static literals to free.
+ */
+void pm_static_literals_free(pm_static_literals_t *literals);
+
+#endif

src/main/c/yarp/include/prism/util/pm_integer.h

@@ -7,6 +7,7 @@
 #define PRISM_NUMBER_H
 
 #include "prism/defines.h"
+#include "prism/util/pm_buffer.h"
 
 #include <assert.h>
 #include <stdbool.h>
@@ -93,6 +94,25 @@ PRISM_EXPORTED_FUNCTION void pm_integer_parse(pm_integer_t *integer, pm_integer_
  */
 size_t pm_integer_memsize(const pm_integer_t *integer);
 
+/**
+ * Compare two integers. This function returns -1 if the left integer is less
+ * than the right integer, 0 if they are equal, and 1 if the left integer is
+ * greater than the right integer.
+ *
+ * @param left The left integer to compare.
+ * @param right The right integer to compare.
+ * @return The result of the comparison.
+ */
+int pm_integer_compare(const pm_integer_t *left, const pm_integer_t *right);
+
+/**
+ * Convert an integer to a decimal string.
+ *
+ * @param buffer The buffer to append the string to.
+ * @param integer The integer to convert to a string.
+ */
+PRISM_EXPORTED_FUNCTION void pm_integer_string(pm_buffer_t *buffer, const pm_integer_t *integer);
+
 /**
  * Free the internal memory of an integer. This memory will only be allocated if
  * the integer exceeds the size of a single node in the linked list.

src/main/c/yarp/src/diagnostic.c

@@ -306,9 +306,12 @@ static const pm_diagnostic_data_t diagnostic_messages[PM_DIAGNOSTIC_ID_LEN] = {
     [PM_WARN_AMBIGUOUS_FIRST_ARGUMENT_PLUS]     = { "ambiguous first argument; put parentheses or a space even after `+` operator", PM_WARNING_LEVEL_VERBOSE },
     [PM_WARN_AMBIGUOUS_PREFIX_STAR]             = { "ambiguous `*` has been interpreted as an argument prefix", PM_WARNING_LEVEL_VERBOSE },
     [PM_WARN_AMBIGUOUS_SLASH]                   = { "ambiguous `/`; wrap regexp in parentheses or add a space after `/` operator", PM_WARNING_LEVEL_VERBOSE },
+    [PM_WARN_DUPLICATED_HASH_KEY]               = { "key %.*s is duplicated and overwritten on line %" PRIi32, PM_WARNING_LEVEL_DEFAULT },
+    [PM_WARN_DUPLICATED_WHEN_CLAUSE]            = { "duplicated 'when' clause with line %" PRIi32 " is ignored", PM_WARNING_LEVEL_VERBOSE },
     [PM_WARN_EQUAL_IN_CONDITIONAL]              = { "found `= literal' in conditional, should be ==", PM_WARNING_LEVEL_DEFAULT },
     [PM_WARN_END_IN_METHOD]                     = { "END in method; use at_exit", PM_WARNING_LEVEL_DEFAULT },
-    [PM_WARN_FLOAT_OUT_OF_RANGE]                = { "Float %.*s%s out of range", PM_WARNING_LEVEL_VERBOSE }
+    [PM_WARN_FLOAT_OUT_OF_RANGE]                = { "Float %.*s%s out of range", PM_WARNING_LEVEL_VERBOSE },
+    [PM_WARN_INTEGER_IN_FLIP_FLOP]              = { "integer literal in flip-flop", PM_WARNING_LEVEL_DEFAULT }
 };
 
 static inline const char *