Handle quoting, escapements and IFS correctly

Signed-off-by: Olivier Gayot <olivier.gayot@sigexec.com>

1 files changed, 98 insertions, 11 deletions

diff --git a/vish.py b/vish.py
index a8febc3..ec601a0 100755
--- a/vish.py
+++ b/vish.py
@@ -4,28 +4,115 @@ import os
 import re
 import argparse
 
+IFS = (" ", "\t", "\n")
+
+
 class EOFException(Exception):
     pass
 
 
+class Argument():
+    def __init__(self, value: str, allow_expansion=True):
+        pass
+
+    def expand():
+        pass
+
+
+class Pipeline():
+    def __init__(self, instructions: list=[]):
+        self.instructions = instructions
+
+    def execute(self):
+        for instruction in self.instructions:
+            instruction.execute()
+
+
 class Instruction():
-    def __init__(self, string: str):
-        # TODO handle quoting and escapments
-        keywords = string.split()
-        self.prog = keywords[0]
-        self.arguments = keywords[1:]
+    def __init__(self, tokens: list=[]):
+        self.prog = tokens[0]
+        self.args = tokens[1:]
 
 
     def execute(self):
         cpid = os.fork()
         if cpid == 0:
-            os.execvp(self.prog, [self.prog] + self.arguments)
+            os.execvp(self.prog, [self.prog] + self.args)
         else:
             os.waitpid(cpid, 0)
 
 
-def read_instruction(fh):
-    # TODO We should be reading an instruction, not a line.
+class PipelineParser():
+    def read_escaped(self, string_iterator):
+        return next(string_iterator)
+
+    def read_literal(self, string_iterator):
+        literal = ""
+        while True:
+            c = next(string_iterator)
+            if c != "'":
+                literal += c
+            else:
+                return literal
+
+    def read_quoted(self, string_iterator):
+        # TODO handle substitutions
+        quoted = ""
+        while True:
+            c = next(string_iterator)
+            if c == "\\":
+                quoted += self.read_escaped(string_iterator)
+            elif c != '"':
+                quoted += c
+            else:
+                return quoted
+
+
+    def get_next_token(self, string_iterator):
+        token = None
+        while True:
+            try:
+                c = next(string_iterator)
+            except StopIteration:
+                break
+
+            # Skip leading whitespaces
+            if token is None and c in IFS:
+                continue
+            elif token is None:
+                token = ""
+
+            if c == "'":
+                token += self.read_literal(string_iterator)
+            elif c == "\\":
+                token += self.read_escaped(string_iterator)
+            elif c == "\"":
+                token += self.read_quoted(string_iterator)
+            elif c in IFS:
+                return token
+            else:
+                token += c
+
+        return token
+
+    def tokenize(self, string_iterator):
+        while True:
+            token = self.get_next_token(string_iterator)
+            if token is None:
+                break
+            yield token
+
+    def parse(self, line):
+        tokens = list(self.tokenize(iter(line)))
+        return Pipeline([Instruction(tokens)])
+
+
+def read_next_pipeline(fh):
+    # TODO Support multiple pipelines per line
+    # TODO Support instructions spawning multiple lines
+
+    parser = PipelineParser()
+
     while True:
         line = fh.readline()
         if line == "":
@@ -36,7 +123,7 @@ def read_instruction(fh):
         if not line.startswith("#") and not line == "":
             break
 
-    return Instruction(line)
+    return parser.parse(line)
 
 
 def main(arguments):
@@ -44,8 +131,8 @@ def main(arguments):
         with open(script, mode="r", encoding="utf-8") as fh:
             while True:
                 try:
-                    instruction = read_instruction(fh)
-                    instruction.execute()
+                    pipeline = read_next_pipeline(fh)
+                    pipeline.execute()
 
                 except EOFException:
                     break