URL
https://opencores.org/ocsvn/openrisc/openrisc/trunk
Subversion Repositories openrisc
[/] [openrisc/] [trunk/] [gnu-dev/] [or1k-gcc/] [libgo/] [go/] [old/] [template/] [parse.go] - Rev 848
Go to most recent revision | Compare with Previous | Blame | View Log
// Copyright 2009 The Go Authors. All rights reserved.
// Use of this source code is governed by a BSD-style
// license that can be found in the LICENSE file.
// Code to parse a template.
package template
import (
"fmt"
"io"
"io/ioutil"
"reflect"
"strconv"
"strings"
"unicode"
"unicode/utf8"
)
// Errors returned during parsing and execution. Users may extract the information and reformat
// if they desire.
type Error struct {
Line int
Msg string
}
func (e *Error) Error() string { return fmt.Sprintf("line %d: %s", e.Line, e.Msg) }
// checkError is a deferred function to turn a panic with type *Error into a plain error return.
// Other panics are unexpected and so are re-enabled.
func checkError(error *error) {
if v := recover(); v != nil {
if e, ok := v.(*Error); ok {
*error = e
} else {
// runtime errors should crash
panic(v)
}
}
}
// Most of the literals are aces.
var lbrace = []byte{'{'}
var rbrace = []byte{'}'}
var space = []byte{' '}
var tab = []byte{'\t'}
// The various types of "tokens", which are plain text or (usually) brace-delimited descriptors
const (
tokAlternates = iota
tokComment
tokEnd
tokLiteral
tokOr
tokRepeated
tokSection
tokText
tokVariable
)
// FormatterMap is the type describing the mapping from formatter
// names to the functions that implement them.
type FormatterMap map[string]func(io.Writer, string, ...interface{})
// Built-in formatters.
var builtins = FormatterMap{
"html": HTMLFormatter,
"str": StringFormatter,
"": StringFormatter,
}
// The parsed state of a template is a vector of xxxElement structs.
// Sections have line numbers so errors can be reported better during execution.
// Plain text.
type textElement struct {
text []byte
}
// A literal such as .meta-left or .meta-right
type literalElement struct {
text []byte
}
// A variable invocation to be evaluated
type variableElement struct {
linenum int
args []interface{} // The fields and literals in the invocation.
fmts []string // Names of formatters to apply. len(fmts) > 0
}
// A variableElement arg to be evaluated as a field name
type fieldName string
// A .section block, possibly with a .or
type sectionElement struct {
linenum int // of .section itself
field string // cursor field for this block
start int // first element
or int // first element of .or block
end int // one beyond last element
}
// A .repeated block, possibly with a .or and a .alternates
type repeatedElement struct {
sectionElement // It has the same structure...
altstart int // ... except for alternates
altend int
}
// Template is the type that represents a template definition.
// It is unchanged after parsing.
type Template struct {
fmap FormatterMap // formatters for variables
// Used during parsing:
ldelim, rdelim []byte // delimiters; default {}
buf []byte // input text to process
p int // position in buf
linenum int // position in input
// Parsed results:
elems []interface{}
}
// New creates a new template with the specified formatter map (which
// may be nil) to define auxiliary functions for formatting variables.
func New(fmap FormatterMap) *Template {
t := new(Template)
t.fmap = fmap
t.ldelim = lbrace
t.rdelim = rbrace
t.elems = make([]interface{}, 0, 16)
return t
}
// Report error and stop executing. The line number must be provided explicitly.
func (t *Template) execError(st *state, line int, err string, args ...interface{}) {
panic(&Error{line, fmt.Sprintf(err, args...)})
}
// Report error, panic to terminate parsing.
// The line number comes from the template state.
func (t *Template) parseError(err string, args ...interface{}) {
panic(&Error{t.linenum, fmt.Sprintf(err, args...)})
}
// Is this an exported - upper case - name?
func isExported(name string) bool {
r, _ := utf8.DecodeRuneInString(name)
return unicode.IsUpper(r)
}
// -- Lexical analysis
// Is c a space character?
func isSpace(c uint8) bool { return c == ' ' || c == '\t' || c == '\r' || c == '\n' }
// Safely, does s[n:n+len(t)] == t?
func equal(s []byte, n int, t []byte) bool {
b := s[n:]
if len(t) > len(b) { // not enough space left for a match.
return false
}
for i, c := range t {
if c != b[i] {
return false
}
}
return true
}
// isQuote returns true if c is a string- or character-delimiting quote character.
func isQuote(c byte) bool {
return c == '"' || c == '`' || c == '\''
}
// endQuote returns the end quote index for the quoted string that
// starts at n, or -1 if no matching end quote is found before the end
// of the line.
func endQuote(s []byte, n int) int {
quote := s[n]
for n++; n < len(s); n++ {
switch s[n] {
case '\\':
if quote == '"' || quote == '\'' {
n++
}
case '\n':
return -1
case quote:
return n
}
}
return -1
}
// nextItem returns the next item from the input buffer. If the returned
// item is empty, we are at EOF. The item will be either a
// delimited string or a non-empty string between delimited
// strings. Tokens stop at (but include, if plain text) a newline.
// Action tokens on a line by themselves drop any space on
// either side, up to and including the newline.
func (t *Template) nextItem() []byte {
startOfLine := t.p == 0 || t.buf[t.p-1] == '\n'
start := t.p
var i int
newline := func() {
t.linenum++
i++
}
// Leading space up to but not including newline
for i = start; i < len(t.buf); i++ {
if t.buf[i] == '\n' || !isSpace(t.buf[i]) {
break
}
}
leadingSpace := i > start
// What's left is nothing, newline, delimited string, or plain text
switch {
case i == len(t.buf):
// EOF; nothing to do
case t.buf[i] == '\n':
newline()
case equal(t.buf, i, t.ldelim):
left := i // Start of left delimiter.
right := -1 // Will be (immediately after) right delimiter.
haveText := false // Delimiters contain text.
i += len(t.ldelim)
// Find the end of the action.
for ; i < len(t.buf); i++ {
if t.buf[i] == '\n' {
break
}
if isQuote(t.buf[i]) {
i = endQuote(t.buf, i)
if i == -1 {
t.parseError("unmatched quote")
return nil
}
continue
}
if equal(t.buf, i, t.rdelim) {
i += len(t.rdelim)
right = i
break
}
haveText = true
}
if right < 0 {
t.parseError("unmatched opening delimiter")
return nil
}
// Is this a special action (starts with '.' or '#') and the only thing on the line?
if startOfLine && haveText {
firstChar := t.buf[left+len(t.ldelim)]
if firstChar == '.' || firstChar == '#' {
// It's special and the first thing on the line. Is it the last?
for j := right; j < len(t.buf) && isSpace(t.buf[j]); j++ {
if t.buf[j] == '\n' {
// Yes it is. Drop the surrounding space and return the {.foo}
t.linenum++
t.p = j + 1
return t.buf[left:right]
}
}
}
}
// No it's not. If there's leading space, return that.
if leadingSpace {
// not trimming space: return leading space if there is some.
t.p = left
return t.buf[start:left]
}
// Return the word, leave the trailing space.
start = left
break
default:
for ; i < len(t.buf); i++ {
if t.buf[i] == '\n' {
newline()
break
}
if equal(t.buf, i, t.ldelim) {
break
}
}
}
item := t.buf[start:i]
t.p = i
return item
}
// Turn a byte array into a space-split array of strings,
// taking into account quoted strings.
func words(buf []byte) []string {
s := make([]string, 0, 5)
for i := 0; i < len(buf); {
// One word per loop
for i < len(buf) && isSpace(buf[i]) {
i++
}
if i == len(buf) {
break
}
// Got a word
start := i
if isQuote(buf[i]) {
i = endQuote(buf, i)
if i < 0 {
i = len(buf)
} else {
i++
}
}
// Even with quotes, break on space only. This handles input
// such as {""|} and catches quoting mistakes.
for i < len(buf) && !isSpace(buf[i]) {
i++
}
s = append(s, string(buf[start:i]))
}
return s
}
// Analyze an item and return its token type and, if it's an action item, an array of
// its constituent words.
func (t *Template) analyze(item []byte) (tok int, w []string) {
// item is known to be non-empty
if !equal(item, 0, t.ldelim) { // doesn't start with left delimiter
tok = tokText
return
}
if !equal(item, len(item)-len(t.rdelim), t.rdelim) { // doesn't end with right delimiter
t.parseError("internal error: unmatched opening delimiter") // lexing should prevent this
return
}
if len(item) <= len(t.ldelim)+len(t.rdelim) { // no contents
t.parseError("empty directive")
return
}
// Comment
if item[len(t.ldelim)] == '#' {
tok = tokComment
return
}
// Split into words
w = words(item[len(t.ldelim) : len(item)-len(t.rdelim)]) // drop final delimiter
if len(w) == 0 {
t.parseError("empty directive")
return
}
first := w[0]
if first[0] != '.' {
tok = tokVariable
return
}
if len(first) > 1 && first[1] >= '0' && first[1] <= '9' {
// Must be a float.
tok = tokVariable
return
}
switch first {
case ".meta-left", ".meta-right", ".space", ".tab":
tok = tokLiteral
return
case ".or":
tok = tokOr
return
case ".end":
tok = tokEnd
return
case ".section":
if len(w) != 2 {
t.parseError("incorrect fields for .section: %s", item)
return
}
tok = tokSection
return
case ".repeated":
if len(w) != 3 || w[1] != "section" {
t.parseError("incorrect fields for .repeated: %s", item)
return
}
tok = tokRepeated
return
case ".alternates":
if len(w) != 2 || w[1] != "with" {
t.parseError("incorrect fields for .alternates: %s", item)
return
}
tok = tokAlternates
return
}
t.parseError("bad directive: %s", item)
return
}
// formatter returns the Formatter with the given name in the Template, or nil if none exists.
func (t *Template) formatter(name string) func(io.Writer, string, ...interface{}) {
if t.fmap != nil {
if fn := t.fmap[name]; fn != nil {
return fn
}
}
return builtins[name]
}
// -- Parsing
// newVariable allocates a new variable-evaluation element.
func (t *Template) newVariable(words []string) *variableElement {
formatters := extractFormatters(words)
args := make([]interface{}, len(words))
// Build argument list, processing any literals
for i, word := range words {
var lerr error
switch word[0] {
case '"', '`', '\'':
v, err := strconv.Unquote(word)
if err == nil && word[0] == '\'' {
args[i], _ = utf8.DecodeRuneInString(v)
} else {
args[i], lerr = v, err
}
case '.', '+', '-', '0', '1', '2', '3', '4', '5', '6', '7', '8', '9':
v, err := strconv.ParseInt(word, 0, 64)
if err == nil {
args[i] = v
} else {
v, err := strconv.ParseFloat(word, 64)
args[i], lerr = v, err
}
default:
args[i] = fieldName(word)
}
if lerr != nil {
t.parseError("invalid literal: %q: %s", word, lerr)
}
}
// We could remember the function address here and avoid the lookup later,
// but it's more dynamic to let the user change the map contents underfoot.
// We do require the name to be present, though.
// Is it in user-supplied map?
for _, f := range formatters {
if t.formatter(f) == nil {
t.parseError("unknown formatter: %q", f)
}
}
return &variableElement{t.linenum, args, formatters}
}
// extractFormatters extracts a list of formatters from words.
// After the final space-separated argument in a variable, formatters may be
// specified separated by pipe symbols. For example: {a b c|d|e}
// The words parameter still has the formatters joined by '|' in the last word.
// extractFormatters splits formatters, replaces the last word with the content
// found before the first '|' within it, and returns the formatters obtained.
// If no formatters are found in words, the default formatter is returned.
func extractFormatters(words []string) (formatters []string) {
// "" is the default formatter.
formatters = []string{""}
if len(words) == 0 {
return
}
var bar int
lastWord := words[len(words)-1]
if isQuote(lastWord[0]) {
end := endQuote([]byte(lastWord), 0)
if end < 0 || end+1 == len(lastWord) || lastWord[end+1] != '|' {
return
}
bar = end + 1
} else {
bar = strings.IndexRune(lastWord, '|')
if bar < 0 {
return
}
}
words[len(words)-1] = lastWord[0:bar]
formatters = strings.Split(lastWord[bar+1:], "|")
return
}
// Grab the next item. If it's simple, just append it to the template.
// Otherwise return its details.
func (t *Template) parseSimple(item []byte) (done bool, tok int, w []string) {
tok, w = t.analyze(item)
done = true // assume for simplicity
switch tok {
case tokComment:
return
case tokText:
t.elems = append(t.elems, &textElement{item})
return
case tokLiteral:
switch w[0] {
case ".meta-left":
t.elems = append(t.elems, &literalElement{t.ldelim})
case ".meta-right":
t.elems = append(t.elems, &literalElement{t.rdelim})
case ".space":
t.elems = append(t.elems, &literalElement{space})
case ".tab":
t.elems = append(t.elems, &literalElement{tab})
default:
t.parseError("internal error: unknown literal: %s", w[0])
}
return
case tokVariable:
t.elems = append(t.elems, t.newVariable(w))
return
}
return false, tok, w
}
// parseRepeated and parseSection are mutually recursive
func (t *Template) parseRepeated(words []string) *repeatedElement {
r := new(repeatedElement)
t.elems = append(t.elems, r)
r.linenum = t.linenum
r.field = words[2]
// Scan section, collecting true and false (.or) blocks.
r.start = len(t.elems)
r.or = -1
r.altstart = -1
r.altend = -1
Loop:
for {
item := t.nextItem()
if len(item) == 0 {
t.parseError("missing .end for .repeated section")
break
}
done, tok, w := t.parseSimple(item)
if done {
continue
}
switch tok {
case tokEnd:
break Loop
case tokOr:
if r.or >= 0 {
t.parseError("extra .or in .repeated section")
break Loop
}
r.altend = len(t.elems)
r.or = len(t.elems)
case tokSection:
t.parseSection(w)
case tokRepeated:
t.parseRepeated(w)
case tokAlternates:
if r.altstart >= 0 {
t.parseError("extra .alternates in .repeated section")
break Loop
}
if r.or >= 0 {
t.parseError(".alternates inside .or block in .repeated section")
break Loop
}
r.altstart = len(t.elems)
default:
t.parseError("internal error: unknown repeated section item: %s", item)
break Loop
}
}
if r.altend < 0 {
r.altend = len(t.elems)
}
r.end = len(t.elems)
return r
}
func (t *Template) parseSection(words []string) *sectionElement {
s := new(sectionElement)
t.elems = append(t.elems, s)
s.linenum = t.linenum
s.field = words[1]
// Scan section, collecting true and false (.or) blocks.
s.start = len(t.elems)
s.or = -1
Loop:
for {
item := t.nextItem()
if len(item) == 0 {
t.parseError("missing .end for .section")
break
}
done, tok, w := t.parseSimple(item)
if done {
continue
}
switch tok {
case tokEnd:
break Loop
case tokOr:
if s.or >= 0 {
t.parseError("extra .or in .section")
break Loop
}
s.or = len(t.elems)
case tokSection:
t.parseSection(w)
case tokRepeated:
t.parseRepeated(w)
case tokAlternates:
t.parseError(".alternates not in .repeated")
default:
t.parseError("internal error: unknown section item: %s", item)
}
}
s.end = len(t.elems)
return s
}
func (t *Template) parse() {
for {
item := t.nextItem()
if len(item) == 0 {
break
}
done, tok, w := t.parseSimple(item)
if done {
continue
}
switch tok {
case tokOr, tokEnd, tokAlternates:
t.parseError("unexpected %s", w[0])
case tokSection:
t.parseSection(w)
case tokRepeated:
t.parseRepeated(w)
default:
t.parseError("internal error: bad directive in parse: %s", item)
}
}
}
// -- Execution
// -- Public interface
// Parse initializes a Template by parsing its definition. The string
// s contains the template text. If any errors occur, Parse returns
// the error.
func (t *Template) Parse(s string) (err error) {
if t.elems == nil {
return &Error{1, "template not allocated with New"}
}
if !validDelim(t.ldelim) || !validDelim(t.rdelim) {
return &Error{1, fmt.Sprintf("bad delimiter strings %q %q", t.ldelim, t.rdelim)}
}
defer checkError(&err)
t.buf = []byte(s)
t.p = 0
t.linenum = 1
t.parse()
return nil
}
// ParseFile is like Parse but reads the template definition from the
// named file.
func (t *Template) ParseFile(filename string) (err error) {
b, err := ioutil.ReadFile(filename)
if err != nil {
return err
}
return t.Parse(string(b))
}
// Execute applies a parsed template to the specified data object,
// generating output to wr.
func (t *Template) Execute(wr io.Writer, data interface{}) (err error) {
// Extract the driver data.
val := reflect.ValueOf(data)
defer checkError(&err)
t.p = 0
t.execute(0, len(t.elems), &state{parent: nil, data: val, wr: wr})
return nil
}
// SetDelims sets the left and right delimiters for operations in the
// template. They are validated during parsing. They could be
// validated here but it's better to keep the routine simple. The
// delimiters are very rarely invalid and Parse has the necessary
// error-handling interface already.
func (t *Template) SetDelims(left, right string) {
t.ldelim = []byte(left)
t.rdelim = []byte(right)
}
// Parse creates a Template with default parameters (such as {} for
// metacharacters). The string s contains the template text while
// the formatter map fmap, which may be nil, defines auxiliary functions
// for formatting variables. The template is returned. If any errors
// occur, err will be non-nil.
func Parse(s string, fmap FormatterMap) (t *Template, err error) {
t = New(fmap)
err = t.Parse(s)
if err != nil {
t = nil
}
return
}
// ParseFile is a wrapper function that creates a Template with default
// parameters (such as {} for metacharacters). The filename identifies
// a file containing the template text, while the formatter map fmap, which
// may be nil, defines auxiliary functions for formatting variables.
// The template is returned. If any errors occur, err will be non-nil.
func ParseFile(filename string, fmap FormatterMap) (t *Template, err error) {
b, err := ioutil.ReadFile(filename)
if err != nil {
return nil, err
}
return Parse(string(b), fmap)
}
// MustParse is like Parse but panics if the template cannot be parsed.
func MustParse(s string, fmap FormatterMap) *Template {
t, err := Parse(s, fmap)
if err != nil {
panic("template.MustParse error: " + err.Error())
}
return t
}
// MustParseFile is like ParseFile but panics if the file cannot be read
// or the template cannot be parsed.
func MustParseFile(filename string, fmap FormatterMap) *Template {
b, err := ioutil.ReadFile(filename)
if err != nil {
panic("template.MustParseFile error: " + err.Error())
}
return MustParse(string(b), fmap)
}
Go to most recent revision | Compare with Previous | Blame | View Log