commit 0a5905e8e6dda95a77cb337806aad4cdbc8116cc
parent a039c7123d3b752e5736add267890a3a64e09997
Author: Byron Torres <b@torresjrjr.com>
Date: Sat, 1 May 2021 20:17:58 +0100
Improve error handling
Diffstat:
| M | nestedtext.go | | | 181 | +++++++++++++++++++++++++++++++++---------------------------------------------- |
1 file changed, 75 insertions(+), 106 deletions(-)
diff --git a/nestedtext.go b/nestedtext.go
@@ -11,6 +11,8 @@ import (
"strings"
)
+// TODO: Impl negative lookahead of ': ' for reKey{Value,Empty}
+
var (
reLeadingWs = regexp.MustCompile(`^\s*`)
reComment = regexp.MustCompile(`^\s*(#.*)?$`)
@@ -18,7 +20,7 @@ var (
reKeyEmpty = regexp.MustCompile(`^\s*[^:]*: ?$`)
reListValue = regexp.MustCompile(`^\s*- .*$`)
reListEmpty = regexp.MustCompile(`^\s*- ?$`)
- reMultiline = regexp.MustCompile(`^\s*> ?.*$`)
+ reMultiline = regexp.MustCompile(`^\s*>( .*)?$`)
)
type BlockType int
@@ -31,16 +33,23 @@ const (
)
func main() {
- DebugDumpJSON(UnmarshelNestedText(os.Stdin))
+ doc, err := UnmarshalNestedText(os.Stdin)
+ must(err)
+ jsonRepr, err := json.MarshalIndent(doc, "", "\t")
+ must(err)
+ fmt.Println(string(jsonRepr))
}
-func UnmarshelNestedText(r io.Reader) interface{} {
+func UnmarshalNestedText(r io.Reader) (interface{}, error) {
scanner := bufio.NewScanner(r)
- return parseNestedText(scanner, "", true)
+ lineNumber := 0
+ return parseNestedText(scanner, "", &lineNumber, true)
}
-func parseNestedText(scanner *bufio.Scanner, parentIndent string, root bool) interface{} {
- blockType := Unknown
+func parseNestedText(
+ scanner *bufio.Scanner, parentIndent string, lineNum *int, root bool,
+) (interface{}, error) {
+ var blockType BlockType = Unknown
var dict map[string]interface{}
var list []interface{}
var multiline string
@@ -54,16 +63,14 @@ func parseNestedText(scanner *bufio.Scanner, parentIndent string, root bool) int
if !scanner.Scan() {
break
}
+ *lineNum++
} else {
skip = false
}
- // fmt.Println()
-
rawLine := scanner.Text()
if reComment.MatchString(rawLine) {
- // fmt.Println("COMMENT, SKIPPING")
continue
}
@@ -71,43 +78,46 @@ func parseNestedText(scanner *bufio.Scanner, parentIndent string, root bool) int
if !localIndentDefined {
localIndent = currentIndent
localIndentDefined = true
- }
-
- // fmt.Printf("parentIndent: |%s|\n", parentIndent)
- // fmt.Printf("localIndent: |%s|\n", localIndent)
- // fmt.Printf("localIndent: |%s|\n", currentIndent)
-
- if (currentIndent != localIndent) ||
+ } else if (currentIndent != localIndent) ||
!strings.HasPrefix(localIndent, parentIndent) {
- // indented leftwards, node has finished.
- // fmt.Println("RETURNING")
- switch blockType {
- case Dict:
- return dict
- case List:
- return list
- case Multiline:
- return multiline
- default:
- return nil
- }
+ // indented leftwards, leaf has finished.
+ break
}
- // match for line types
+ // the effective current line
line := strings.TrimPrefix(rawLine, localIndent)
- // fmt.Println("DEBUG: rawLine |" + rawLine)
- // fmt.Println("DEBUG: line |" + line)
-
- if reKeyValue.MatchString(line) {
+ // determine the blocktype from the current linetype
+ errmsg := "ERROR: inconsistent linetypes - line %d: \"%s\""
+ switch {
+ case reKeyValue.MatchString(line) || reKeyEmpty.MatchString(line):
if blockType == Unknown {
blockType = Dict
dict = make(map[string]interface{})
} else if blockType != Dict {
- // fmt.Println("ERROR: mixed line types in block")
- return dict
+ err := fmt.Errorf(errmsg, *lineNum, line)
+ return nil, err
+ }
+ case reListValue.MatchString(line) || reListEmpty.MatchString(line):
+ if blockType == Unknown {
+ blockType = List
+ list = []interface{}{}
+ } else if blockType != List {
+ err := fmt.Errorf(errmsg, *lineNum, line)
+ return nil, err
+ }
+ case reMultiline.MatchString(line):
+ if blockType == Unknown {
+ blockType = Multiline
+ } else if blockType != Multiline {
+ err := fmt.Errorf(errmsg, *lineNum, line)
+ return nil, err
}
+ }
+ // parse line according to linetype, then continue to next cycle
+ switch {
+ case reKeyValue.MatchString(line):
// TODO optimise with regex?
ls := strings.Split(line, ": ")
key, v := ls[0], ls[1:]
@@ -115,108 +125,67 @@ func parseNestedText(scanner *bufio.Scanner, parentIndent string, root bool) int
value := strings.Join(v, ": ")
dict[key] = value
- // DebugDumpJSON(dict)
- continue
- }
- if reKeyEmpty.MatchString(line) {
- if blockType == Unknown {
- blockType = Dict
- dict = make(map[string]interface{})
- } else if blockType != Dict {
- // fmt.Println("ERROR: mixed line types in block")
- return dict
+ case reKeyEmpty.MatchString(line):
+ var key string
+ switch {
+ case strings.HasSuffix(line, ":"):
+ key = strings.TrimSuffix(line, ":")
+ case strings.HasSuffix(line, ": "):
+ key = strings.TrimSuffix(line, ": ")
}
-
- // fmt.Println("NESTING")
- // TODO: Check if correct.
- key := strings.TrimSuffix(line, ":")
- key = strings.TrimSuffix(key, ": ")
key = strings.TrimSpace(key)
- child := parseNestedText(scanner, localIndent, false)
+ child, err := parseNestedText(scanner, localIndent, lineNum, false)
+ if err != nil {
+ return nil, err
+ }
dict[key] = child
-
- // when parseNestedText returns, scanner will be
- // at the first line of the nest block.
skip = true
- // DebugDumpJSON(dict)
- continue
- }
- if reListValue.MatchString(line) {
- if blockType == Unknown {
- blockType = List
- list = []interface{}{}
- } else if blockType != List {
- // fmt.Println("ERROR: mixed line types in block")
- return list
- }
-
+ case reListValue.MatchString(line):
value := strings.TrimPrefix(line, "- ")
list = append(list, value)
- // DebugDumpJSON(list)
- continue
- }
- if reListEmpty.MatchString(line) {
- if blockType == Unknown {
- blockType = List
- list = []interface{}{}
- } else if blockType != List {
- // fmt.Println("ERROR: mixed line types in block")
- return list
+ case reListEmpty.MatchString(line):
+ child, err := parseNestedText(scanner, localIndent, lineNum, false)
+ if err != nil {
+ return nil, err
}
-
- // fmt.Println("NESTING")
-
- child := parseNestedText(scanner, localIndent, false)
list = append(list, child)
-
- // when parseNestedText returns, scanner will be
- // at the first line of the nest block.
skip = true
- // DebugDumpJSON(dict)
- continue
- }
- if reMultiline.MatchString(line) {
- if blockType == Unknown {
- blockType = Multiline
- } else if blockType != Multiline {
- // fmt.Println("ERROR: mixed line types in block")
- return multiline
- }
-
- oneline := ""
+ case reMultiline.MatchString(line):
+ var oneline string
if line != ">" {
oneline = strings.TrimPrefix(line, "> ")
}
multiline += oneline + "\n"
- // DebugDumpJSON(multiline)
- continue
+ default:
+ errmsg := "Internal error - unknown linetype of line %d"
+ return nil, fmt.Errorf(errmsg, &lineNum)
}
}
- // fmt.Println("SCAN FINISHED")
-
if err := scanner.Err(); err != nil {
log.Println(err)
}
switch blockType {
case Dict:
- return dict
+ return dict, nil
case List:
- return list
+ return list, nil
case Multiline:
- return multiline
+ return multiline, nil
default:
- return nil
+ return nil, fmt.Errorf("Internal error - line %d", &lineNum)
}
}
-func DebugDumpJSON(doc interface{}) {
- jsonRepr, _ := json.MarshalIndent(doc, "", " ")
- fmt.Println(string(jsonRepr))
+func must(err error) {
+ if err != nil {
+ fmt.Fprintln(os.Stderr, err)
+ os.Exit(1)
+ }
}
