TextLooksImpl.mesa
Copyright © 1985, 1986 by Xerox Corporation.  All rights reserved.
written by Bill Paxton, February 1981
Bill Paxton, December 13, 1982 1:17 pm
Maxwell, January 5, 1983 3:54 pm
Russ Atkinson, July 25, 1983 3:36 pm
Michael Plass, March 29, 1985 5:24:46 pm PST
Doug Wyatt, September 4, 1986 6:09:19 pm PDT
 
DIRECTORY
Basics USING [DoubleAnd, DoubleNot, DoubleOr, NonNegative], 
PrincOpsUtils USING [LongCopy], 
Rope USING [Cat, Fetch, FromChar, ROPE, Size], 
RunReader USING [FreeRunReader, Get, GetRunReader, Ref, SetPosition], 
TextLooks USING [allLooks, BaseRuns, FlatMax, Look, Looks, MaxLen, noLooks, Run, Runs, RunsBody, Tchange, Tconcat, Treplace, Tsubstr], 
TextLooksSupport USING [];
 
TextLooksBasicImpl
ReplaceByRun: 
PUBLIC 
PROC [dest: Runs, start, len, runLen, destSize: 
INT, inherit: 
BOOL, looks: Looks] 
RETURNS [Runs] = {
merge: BOOL ← FALSE;
mergeLooks: Looks;
split, numruns: NAT;
flat: BaseRuns;
c, numRuns, oldPos, size: INT;
Count: 
PROC [start, len: 
INT] 
RETURNS [
INT] = {
IF len=0 THEN RETURN[numRuns];
[c, merge, mergeLooks] ←
CountRuns[dest, start, len, FlatMax-numRuns, merge, mergeLooks];
RETURN [numRuns←numRuns+c];
};
 
AddIt: 
PROC 
RETURNS [
INT] = {
c ← IF merge AND mergeLooks=looks THEN 0 ELSE 1;
merge ← TRUE; mergeLooks ← looks;
RETURN [numRuns←numRuns+c];
};
 
Extract: 
PROC [start, len: 
INT] = {
IF len > 0 THEN split ← ExtractRuns[flat, dest, start, len, split];
};
 
TryFlatAppendRun: 
PROC [base: Runs] 
RETURNS [Runs] = {
flat: BaseRuns;
size: INT;
[numRuns, merge, mergeLooks] ← CountRuns[base, 0, size←Size[base], FlatMax];
IF numRuns > FlatMax OR AddIt[] > FlatMax THEN RETURN [NIL];
flat ← NewBase[numruns←numRuns];
split ← ExtractRuns[flat, base, 0, size, 0];
split ← InsertRun[flat, runLen, looks, split];
IF split # numruns THEN ERROR;
IF flat[numruns-1].after # size+runLen THEN ERROR;
RETURN [flat];
};
 
IF runLen=0 THEN RETURN [Replace[base: dest, start: start, len: len, replace: NIL, baseSize: destSize, repSize: 0, tryFlat: TRUE]];
IF inherit 
THEN {
-- get looks from destination
IF destSize = 0 THEN NULL -- take from arg list
ELSE looks ← FetchLooks[dest, 
IF start > 0 THEN start-1 -- take from before replacement
ELSE IF len=destSize THEN 0 -- replacing everything
ELSE len]; -- take from after replacement
 
};
 
IF dest=NIL AND looks=noLooks THEN RETURN[NIL];
numRuns ← 0; oldPos ← start+len; size ← destSize-len+runLen;
IF Count[0, start] > FlatMax 
OR AddIt[] > FlatMax 
OR Count[oldPos, destSize-oldPos] > FlatMax 
THEN {
newPos: INT ← start+runLen;
replace, new: Runs;
WHILE dest # 
NIL 
DO
WITH dest 
SELECT 
FROM
x: 
REF RunsBody.node.replace => {
xnewPos: INT ← x.newPos;
xstart: INT ← x.start;
IF start <= xstart 
AND oldPos >= xnewPos 
THEN {
replacing the replacement
oldPos ← x.oldPos+(oldPos-xnewPos); dest ← x.base; len ← oldPos-start; LOOP;
}
 
ELSE IF start = xnewPos -- appending to the replacement
AND (new ← TryFlatAppendRun[x.replace])#
NIL 
THEN {
start ← xstart; oldPos ← x.oldPos+len;
dest ← x.base; replace ← new;
Exit the loop from here; len and runLen are not needed anymore, so don't upate the values. 
}
 
};
x: 
REF RunsBody.node.concat => { 
-- try to append to first part of the concat
xpos: INT ← x.pos;
IF start=xpos AND len=0 AND 
(new ← TryFlatAppendRun[x.base])#
NIL 
THEN 
RETURN [
NEW[Tconcat ← [node[concat
[size, new, x.rest, xpos+runLen]]]]];
 
};
ENDCASE;
 
EXIT;
ENDLOOP;
 
IF replace=NIL AND (replace ← CreateRun[runLen, looks])=NIL THEN
replace ← MakeRun[runLen];
IF dest=NIL THEN dest ← MakeRun[destSize];
RETURN [NEW[Treplace ← [node[replace[size, dest, replace, start, oldPos, newPos]]]]]
};
 
IF numRuns=0 THEN RETURN[NIL];
flat ← NewBase[numruns←numRuns]; split ← 0;
Extract[0, start];
split ← InsertRun[flat, runLen, looks, split];
Extract[oldPos, destSize-oldPos];
IF split # numruns THEN ERROR;
IF flat[numruns-1].after # size THEN ERROR;
RETURN [flat] 
};
 
CountRuns: 
PUBLIC 
PROC [runs: Runs, start, len: 
INT, limit: 
INT ← MaxLen, merge: 
BOOL ← 
FALSE, firstLooks: Looks ← noLooks] 
RETURNS [count: 
INT, nonempty: 
BOOL, lastLooks: Looks] = {
stops counting when exceeds limit
if merge is true, then doesn't count first run if its looks=firstLooks
c: INT;
count ← 0;
DO nonempty ← merge;
IF len=0 THEN { lastLooks ← firstLooks; RETURN };
IF runs=
NIL 
THEN {
c ← IF merge AND firstLooks=noLooks THEN 0 ELSE 1;
RETURN [count+c, TRUE, noLooks] 
};
 
WITH runs 
SELECT 
FROM
x: 
REF RunsBody.base => {
first, last: NAT;
len ← MIN[len, CheckLongSub[TbaseSize[x], start]];
[first, last] ← FindBaseRuns[x, start, len];
c ← last-first+1;
IF merge AND firstLooks=x[first].looks THEN c ← c-1;
RETURN [count+c, TRUE, x[last].looks]
};
x: 
REF RunsBody.node.substr => {
len ← MIN[len, CheckLongSub[x.size, start]];
start ← start + x.start; runs ← x.base; LOOP
};
x: 
REF RunsBody.node.concat => {
xpos: INT ← x.pos;
len ← MIN[len, CheckLongSub[x.size, start]];
IF start < xpos 
THEN {
subLen: INT ← xpos - start;
IF len <= subLen THEN { runs ← x.base; LOOP };
[c, merge, firstLooks] ← CountRuns[
x.base, start, subLen, limit, merge, firstLooks]; 
count ← count+c; IF c > limit THEN RETURN;
limit ← limit-c; start ← xpos; len ← len-subLen;
};
 
start ← start-xpos; runs ← x.rest; LOOP 
};
x: 
REF RunsBody.node.replace => {
xstart: INT ← x.start;
xnew: INT ← x.newPos;
len ← MIN[len, CheckLongSub[x.size, start]];
IF start < xstart 
THEN {
subLen: INT ← xstart - start;
IF len <= subLen THEN {runs ← x.base; LOOP};
[c, merge, firstLooks] ← CountRuns[
x.base, start, subLen, limit, merge, firstLooks]; 
count ← count+c; IF c > limit THEN RETURN;
limit ← limit-c;
start ← xstart; len ← len-subLen;
};
 
IF start < xnew 
THEN {
st: INT ← start - xstart;
subLen: INT ← xnew - start;
IF len <= subLen 
THEN {
start ← st; runs ← x.replace; LOOP
};
 
[c, merge, firstLooks] ← CountRuns[
x.replace, st, subLen, limit, merge, firstLooks]; 
count ← count+c; IF c > limit THEN RETURN;
limit ← limit-c; start ← xnew; len ← len-subLen
};
 
start ← start - xnew + x.oldPos; runs ← x.base; LOOP
};
x: 
REF RunsBody.node.change => {
xstart: INT ← x.start;
xend, subLen: INT;
len ← MIN[len, CheckLongSub[x.size, start]];
IF start < xstart 
THEN {
IF len <= (subLen ← xstart-start) THEN {runs ← x.base; LOOP};
[c, merge, firstLooks] ← CountRuns[
x.base, start, subLen, limit, merge, firstLooks]; 
count ← count+c; IF c > limit THEN RETURN;
limit ← limit-c;
start ← xstart; len ← len-subLen;
};
 
IF start < (xend ← xstart+x.len) 
THEN {
subLen ← MIN[xend-start, len];
[c, merge, firstLooks] ← CountRunsAfterChanges[
x.base, start, subLen, limit, 
x.remove, x.add, merge, firstLooks]; 
count ← count+c; IF c > limit THEN RETURN;
limit ← limit-c; start ← xend; len ← len-subLen;
};
 
runs ← x.base; LOOP
};
ENDCASE => ERROR;
 
ENDLOOP;
 
};
 
ExtractRuns: 
PUBLIC 
PROC [base: BaseRuns, ref: Runs, start, len: 
INT, index: 
NAT ← 0] 
RETURNS [
NAT] = 
TRUSTED { 
-- value is next index
DO 
IF len=0 THEN RETURN [index];
IF ref=NIL THEN -- treat as noLooks
RETURN [InsertRun[base, len, noLooks, index]];
WITH ref 
SELECT 
FROM
x: 
REF RunsBody.base => {
firstLen, lastLen, xloc, next, loc: INT;
first, last: NAT;
len ← MIN[len, CheckLongSub[TbaseSize[x], start]];
[first, last] ← FindBaseRuns[x, start, len];
[firstLen, lastLen] ← BaseRunLengths[x, start, len, first, last];
now extract the runs
IF index=0 
THEN { 
-- this is the first run to be extracted
loc ← firstLen; base[0] ← [loc, x[first].looks]; index ← 1;
}
 
ELSE {
loc ← base[index-1].after + firstLen;
IF base[index-1].looks=x[first].looks -- merge runs
THEN base[index-1].after ← loc
ELSE { base[index] ← [loc, x[first].looks]; index ← index+1 }
 
};
 
IF first=last THEN RETURN [index];
IF (xloc ← x[first].after) = loc 
THEN { 
-- can simply copy runs
numRuns: NAT;
IF (numRuns ← last-first-1) > 0 
THEN {
CopyRuns[to:base, toLoc:index, 
from:x, fromLoc:first+1, nRuns: numRuns];
index ← index+numRuns; loc ← base[index-1].after 
}
 
}
 
ELSE 
FOR i: 
NAT 
IN (first..last) 
DO
loc ← loc + (next ← x[i].after) - xloc; xloc ← next;
base[index] ← [loc, x[i].looks];
index ← index+1; ENDLOOP;
 
base[index] ←  [loc+lastLen, x[last].looks];
RETURN [index+1] 
};
x: 
REF RunsBody.node.substr => {
len ← MIN[len, CheckLongSub[x.size, start]];
start ← start + x.start; ref ← x.base; LOOP
};
x: 
REF RunsBody.node.concat => {
xpos: INT ← x.pos;
len ← MIN[len, CheckLongSub[x.size, start]];
IF start < xpos 
THEN {
subLen: INT ← xpos - start;
IF len <= subLen THEN { ref ← x.base; LOOP };
index ← ExtractRuns[base, x.base, start, subLen, index];
start ← xpos; len ← len-subLen 
}; 
 
start ← start-xpos; ref ← x.rest; LOOP 
};
x: 
REF RunsBody.node.replace => {
xstart: INT ← x.start;
xnew: INT ← x.newPos;
len ← MIN[len, CheckLongSub[x.size, start]];
IF start < xstart 
THEN {
subLen: INT ← xstart - start;
IF len <= subLen THEN {ref ← x.base; LOOP};
index ← ExtractRuns[base, x.base, start, subLen, index]; 
start ← xstart; len ← len-subLen
};
 
IF start < xnew 
THEN {
st: INT ← start - xstart;
subLen: INT ← xnew - start;
IF len <= subLen 
THEN {
start ← st; ref ← x.replace; LOOP
};
 
index ← ExtractRuns[base, x.replace, st, subLen, index]; 
start ← xnew; len ← len-subLen
};
 
start ← start - xnew + x.oldPos; ref ← x.base; LOOP
};
x: 
REF RunsBody.node.change => {
xstart: INT ← x.start;
xend, subLen: INT;
len ← MIN[len, CheckLongSub[x.size, start]];
IF start < xstart 
THEN {
IF len <= (subLen ← xstart-start) THEN {ref ← x.base; LOOP};
index ← ExtractRuns[base, x.base, start, subLen, index]; 
start ← xstart; len ← len-subLen
};
 
IF start < (xend ← xstart+x.len) 
THEN {
subLen ← MIN[xend-start, len];
index ← ExtractRunsAfterChanges[
base, x.base, x.remove, x.add, start, subLen, index]; 
start ← xend; len ← len-subLen
};
 
ref ← x.base; LOOP
};
ENDCASE => ERROR;
 
ENDLOOP
 
};
 
NewBase: 
PUBLIC 
PROC [runs: 
NAT] 
RETURNS [BaseRuns] = {
RETURN [NEW[base RunsBody[runs]]];
};
 
BaseRun: 
PUBLIC 
PROC [x: BaseRuns, index: 
INT, lower: 
NAT ← 0, upper: 
NAT ← 
LAST[
NAT]] 
RETURNS [
NAT] = {
len: NAT;
size: INT;
IF index = 0 THEN RETURN [0];
IF (len←x.length) <= 1 THEN RETURN [0];
IF index+1 >= (size←TbaseSize[x]) THEN RETURN[len-1];
IF upper >= len THEN upper ← len-1;
IF lower > 0 AND x[lower-1].after > index THEN lower ← 0;
DO 
-- always know index is in run between lower and upper inclusive
run: NAT ← (upper+lower)/2;
IF index < x[run].after THEN upper ← run ELSE lower ← run+1;
IF upper=lower THEN RETURN[upper];
ENDLOOP 
 
};
 
CopyRuns: 
PUBLIC 
PROC [to, from: BaseRuns, toLoc, fromLoc, nRuns: 
NAT] = 
TRUSTED {
RunsOffset: NAT = SIZE[base RunsBody[0]];
nLeft: NAT;
IF nRuns=0 THEN RETURN;
IF fromLoc >= from.length OR toLoc >= to.length THEN ERROR;
nLeft ← nRuns-1;
to[toLoc+nLeft] ← from[fromLoc+nLeft]; -- bounds check
PrincOpsUtils.LongCopy[
from: LOOPHOLE[from, LONG POINTER]+fromLoc*SIZE[Run]+RunsOffset, 
to: LOOPHOLE[to, LONG POINTER]+toLoc*SIZE[Run]+RunsOffset, 
nwords: nLeft*SIZE[Run]]
};
 
InsertRun: 
PUBLIC 
PROC [base: BaseRuns, len: 
INT, looks: Looks, index: 
NAT] 
RETURNS [
NAT] = { 
-- value is next index
IF index=0 THEN { base[0] ← [len, looks]; index ← 1 }
ELSE {
loc: INT ← base[index-1].after + len;
IF base[index-1].looks=looks -- merge runs
THEN base[index-1].after ← loc
ELSE { base[index] ← [loc, looks]; index ← index+1 }
 
};
 
RETURN [index]
};
 
FindBaseRuns: 
PUBLIC 
PROC [x: BaseRuns, start, len: 
INT] 
RETURNS [first, last: 
NAT] = {
first ← BaseRun[x, start];
last ← 
IF len>1 
THEN BaseRun[x, start+len-1, first] 
ELSE first 
};
 
BaseRunLengths: 
PUBLIC 
PROC [x: BaseRuns, start, len: 
INT, first, last: 
NAT] 
RETURNS [firstLen, lastLen: 
INT] = {
IF first=last THEN RETURN[len, len];
RETURN[x[first].after-start, start+len-x[last-1].after] 
};
 
And: 
PROC [a, b: Looks] 
RETURNS [Looks] ~ 
INLINE {
RETURN[LOOPHOLE[Basics.DoubleAnd[LOOPHOLE[a], LOOPHOLE[b]]]];
};
 
Or: 
PROC [a, b: Looks] 
RETURNS [Looks] ~ 
INLINE {
RETURN[LOOPHOLE[Basics.DoubleOr[LOOPHOLE[a], LOOPHOLE[b]]]];
};
 
Not: 
PROC [a: Looks] 
RETURNS [Looks] ~ 
INLINE {
RETURN[LOOPHOLE[Basics.DoubleNot[LOOPHOLE[a]]]];
};
 
LooksAND: 
PUBLIC 
PROC [looks1, looks2: Looks] 
RETURNS [Looks] ~ {
RETURN[And[looks1, looks2]];
};
 
LooksOR: 
PUBLIC 
PROC [looks1, looks2: Looks] 
RETURNS [Looks] ~ {
RETURN[Or[looks1, looks2]];
};
 
LooksNOT: 
PUBLIC 
PROC [looks: Looks] 
RETURNS [Looks] ~ {
RETURN[Not[looks]];
};
 
ModifyLooks: 
PUBLIC 
PROC [old, remove, add: Looks] 
RETURNS [Looks] = {
modified looks are == (old & ~remove) v add
RETURN[Or[And[old, Not[remove]], add]];
};
 
MergeChanges: 
PUBLIC 
PROC [oldrem, oldadd, rem, add: Looks] 
RETURNS [newrem, newadd: Looks] = { 
((lks & ~oldrem) v oldadd) & ~rem) v add ==
lks & ~(oldrem v rem)) v ((oldadd & ~rem) v add
thus, newrem ← oldrem v rem, newadd ← (oldadd & ~rem) v add
RETURN[newrem: Or[oldrem, rem], newadd: Or[And[oldadd, Not[rem]], add]];
};
 
 
Edit operations
Size: 
PUBLIC 
PROC [base: Runs] 
RETURNS [size: 
INT] = {
RETURN [
IF base = 
NIL 
THEN 0 
ELSE 
WITH base 
SELECT 
FROM
x: REF RunsBody.base => IF x.length=0 THEN 0 ELSE x[x.length-1].after, 
x: REF RunsBody.node.substr => x.size, 
x: REF RunsBody.node.concat => x.size, 
x: REF RunsBody.node.replace => x.size, 
x: REF RunsBody.node.change => x.size, 
ENDCASE => ERROR
]
 
};
 
Substr: PUBLIC PROC [base: Runs, start: INT, len: INT]
RETURNS [new: Runs] = {
DO
IF base=NIL OR len=0 THEN RETURN[NIL]; 
WITH base 
SELECT 
FROM
x: 
REF RunsBody.base => {
rem: INT;
IF (rem ← TbaseSize[x]-start) <= len THEN
IF start = 0 THEN RETURN [base] ELSE len ← rem;
};
x: 
REF RunsBody.node.substr => {
rem: INT;
IF (rem ← CheckLongSub[x.size, start]) <= len THEN
IF start = 0 THEN RETURN [base] ELSE len ← rem;
start ← start + x.start; base ← x.base; LOOP;
};
x: 
REF RunsBody.node.concat => {
xpos, rem: INT;
IF (rem ← CheckLongSub[x.size, start]) <= len THEN
IF start = 0 THEN RETURN [base] ELSE len ← rem;
IF start >= (xpos ← x.pos) 
THEN {
start ← start - xpos; base ← x.rest; LOOP
};
 
IF xpos >= start+len THEN {base ← x.base; LOOP};
};
x: 
REF RunsBody.node.replace => {
xstart, xnew, rem: INT;
IF (rem ← CheckLongSub[x.size, start]) <= len THEN
IF start = 0 THEN RETURN [base] ELSE len ← rem;
IF start >= (xnew ← x.newPos) 
THEN {
start ← start - xnew + x.oldPos; base ← x.base; LOOP
};
 
IF (rem ← start+len) <= (xstart ← x.start) 
THEN {
base ← x.base; LOOP
};
 
IF start >= xstart 
AND rem <= xnew 
THEN {
start ← start - xstart; base ← x.replace; LOOP
};
 
};
x: 
REF RunsBody.node.change => {
xstart: INT ← x.start;
xend: INT ← xstart+x.len;
IF start >= xend 
OR start+len <= xstart 
THEN {
base ← x.base; LOOP
};
 
};
ENDCASE => ERROR;
 
IF (new ← TryFlatSubstr[base, start, len]) # NIL THEN RETURN;
EXIT;
ENDLOOP;
 
IF base=NIL THEN ERROR;
RETURN [NEW[Tsubstr ← [node[substr[len, base, start]]]]];
};
 
TryFlatSubstr: PUBLIC PROC
[base: Runs, start, len: 
INT, limit: 
INT ← FlatMax]
RETURNS [BaseRuns] = { -- return NIL if couldn't flatten
count: INT;
numruns: NAT;
flat: BaseRuns;
[count, , ] ← CountRuns[base, start, len, limit];
IF count > limit THEN RETURN [NIL];
flat ← NewBase[numruns𡤌ount];
IF ExtractRuns[flat, base, start, len] # numruns THEN ERROR;
IF flat[numruns-1].after # len THEN ERROR;
RETURN [flat];
};
Flatten: 
PUBLIC 
PROC [base: Runs] 
RETURNS [new: Runs] = {
size, half: INT;
IF base=NIL THEN RETURN [NIL];
WITH base 
SELECT 
FROM
x: REF RunsBody.base => RETURN [x]; -- already flat
ENDCASE => NULL;
 
new ← TryFlatSubstr[base, 0, Size[base], 8000];
IF new#NIL THEN RETURN; -- else was too big to flatten in one piece
half ← (size ← Size[base])/2; -- flatten the halves
RETURN [Concat[
Flatten[Substr[base, 0, half]], 
Flatten[Substr[base, half, size]], 
half, 
size-half]];
 
};
 
Concat: PUBLIC PROC
[base, rest: Runs, baseLen, restLen: 
INT] 
RETURNS [new: Runs] = 
TRUSTED {
c, numRuns, size, flatLen: INT;
split: NAT;
merge: BOOL;
looks: Looks;
IF base=NIL AND rest=NIL THEN RETURN[NIL];
IF restLen=0 THEN RETURN [base];
IF baseLen=0 THEN RETURN [rest];
size ← baseLen+restLen;
[numRuns, merge, looks] ← CountRuns[base, 0, baseLen, FlatMax];
IF numRuns <= FlatMax 
THEN { 
-- try flattening base
IF (new ← TryFlatConcatRest[base, rest, baseLen, restLen, 
numRuns, merge, looks]) # NIL THEN RETURN;
 
otherwise try to break up rest
IF rest # 
NIL 
THEN 
WITH rest 
SELECT 
FROM
x: 
REF RunsBody.node.concat => { 
-- try to combine base & rest.base
[c, , ] ← CountRuns[x.base, 0, x.pos, FlatMax-numRuns, merge, looks];
IF (numRuns ← numRuns+c) <= FlatMax 
THEN { 
-- flatten
numruns: NAT;
flat: BaseRuns ← NewBase[numruns←numRuns];
split ← ExtractRuns[flat, base, 0, baseLen];
IF ExtractRuns[flat, x.base, 0, x.pos, split] # numruns
THEN ERROR;
flatLen ← baseLen+x.pos;
IF flat[numruns-1].after # flatLen THEN ERROR;
RETURN[
NEW[Tconcat ← [node[concat
[size, flat, x.rest, flatLen]]]]]
 
}
 
};
ENDCASE => NULL;
 
};
 
IF base # 
NIL 
THEN 
WITH base 
SELECT 
FROM
x: 
REF RunsBody.node.concat => { 
-- try to combine base.rest & rest
baseRestLen: INT ← baseLen-x.pos;
[numRuns, merge, looks] ← CountRuns[x.rest, 0, baseRestLen, FlatMax];
IF numRuns <= FlatMax 
THEN {
[c, , ] ← CountRuns[rest, 0, restLen, FlatMax-numRuns, merge, looks];
IF (numRuns ← numRuns+c) <= FlatMax 
THEN { 
-- flatten
numruns: NAT;
flat: BaseRuns ← NewBase[numruns←numRuns];
split ← ExtractRuns[flat, x.rest, 0, baseRestLen];
IF ExtractRuns[flat, rest, 0, restLen, split] # numruns
THEN ERROR;
IF flat[numruns-1].after # baseRestLen+restLen THEN ERROR;
RETURN[
NEW[Tconcat ← [node[concat
[size, x.base, flat, x.pos]]]]]
 
}
 
}
 
};
x: 
REF RunsBody.node.substr => {
see if doing concat of adjacent substr's
WITH rest 
SELECT 
FROM
y: 
REF RunsBody.node.substr => {
see if adjacent to x
IF x.base = y.base AND x.start+x.size = y.start THEN -- join them
RETURN[
NEW[Tsubstr ←
[node[substr[size, x.base, x.start]]]]];
 
};
ENDCASE => NULL;
 
};
ENDCASE => NULL;
 
IF base=NIL THEN base ← MakeRun[baseLen];
IF rest=NIL THEN rest ← MakeRun[restLen];
RETURN[NEW[Tconcat ← [node[concat[size, base, rest, baseLen]]]]];
};
TryFlatConcat: PUBLIC PROC
[base, rest: Runs, baseLen, restLen: INT]
RETURNS [new: BaseRuns] = { 
-- returns NIL if too big to flatten
numRuns: INT;
merge: BOOL;
looks: Looks;
[numRuns, merge, looks] ← CountRuns[base, 0, baseLen, FlatMax];
IF numRuns > FlatMax THEN RETURN [NIL];
RETURN [TryFlatConcatRest[base, rest, baseLen, restLen, numRuns, merge, looks]];
};
 
TryFlatConcatRest: PUBLIC PROC
[base, rest: Runs, baseLen, restLen, numRuns: 
INT, 
merge: BOOL, looks: Looks]
RETURNS [BaseRuns] = { 
-- returns NIL if too big to flatten
c: INT;
split, numruns: NAT;
flat: BaseRuns;
[c, , ] ← CountRuns[rest, 0, restLen, FlatMax-numRuns, merge, looks];
IF (numRuns ← numRuns+c) > FlatMax THEN RETURN [NIL];
flat ← NewBase[numruns←numRuns];
split ← ExtractRuns[flat, base, 0, baseLen];
IF ExtractRuns[flat, rest, 0, restLen, split] # numruns THEN
ERROR;
IF flat[numruns-1].after # baseLen+restLen THEN ERROR;
RETURN [flat];
};
 
Replace: 
PUBLIC 
PROC [
base: Runs, start, len: INT, replace: Runs, 
baseSize, repSize: INT, tryFlat: BOOL ← TRUE]
RETURNS [new: Runs] = {
oldPos, newPos, size: INT;
IF base=NIL AND replace=NIL THEN RETURN [NIL];
oldPos ← start+len;
newPos ← start+repSize;
size ← baseSize-len+repSize;
IF repSize=0 THEN -- deleting
IF base=NIL OR (start=0 AND len=baseSize) THEN RETURN[NIL]
ELSE IF len=0 THEN RETURN[base];
IF tryFlat 
THEN {
merge: BOOL ← FALSE;
flat: BaseRuns;
looks: Looks;
c, numRuns: INT;
split, numruns: NAT;
Count: 
PROC [r: Runs, start, len: 
INT] 
RETURNS [
INT] = 
TRUSTED {
IF len=0 THEN RETURN[numRuns];
[c, merge, looks] ← CountRuns[r, start, len, FlatMax-numRuns, merge, looks];
RETURN [numRuns←numRuns+c]
};
 
Extract: 
PROC [r: Runs, start, len: 
INT] = 
TRUSTED {
IF len > 0 THEN split ← ExtractRuns[flat, r, start, len, split] 
};
 
numRuns ← 0;
IF Count[base, 0, start] <= FlatMax AND
Count[replace, 0, repSize] <= FlatMax AND
Count[base, oldPos, baseSize-oldPos] <= FlatMax 
THEN {
flat ← NewBase[numruns←numRuns]; split ← 0;
Extract[base, 0, start]; Extract[replace, 0, repSize];
Extract[base, oldPos, baseSize-oldPos];
IF split # numruns THEN ERROR;
IF flat[numruns-1].after # size THEN ERROR;
RETURN [flat] 
}
};
 
WHILE base # 
NIL 
DO 
WITH base 
SELECT 
FROM
x: 
REF RunsBody.node.replace => {
xnewPos: INT ← x.newPos;
xstart: INT ← x.start;
xsize: INT;
IF start <= xstart 
AND oldPos >= xnewPos 
THEN {
replacing the replacement
oldPos ← x.oldPos+(oldPos-xnewPos); len ← oldPos-start;
base ← x.base; LOOP
}
 
ELSE IF repSize = 0 AND start > xstart AND
oldPos = xnewPos AND -- deleting end of prior replacement
(new ← TryFlatSubstr[x.replace, 0, start-xstart])#
NIL 
THEN {
newPos ← start; oldPos ← x.oldPos; start ← xstart;
replace ← new; base ← x.base
}
ELSE IF repSize = 0 AND start = xstart AND
oldPos < xnewPos AND -- deleting start of prior replacement
(new ← TryFlatSubstr[x.replace, len, xnewPos-oldPos])#
NIL 
THEN {
newPos ← start+xnewPos-oldPos; oldPos ← x.oldPos;
replace ← new; base ← x.base
}
ELSE IF start = xnewPos AND -- replacing just after prior replacement
(new ← TryFlatConcat[x.replace, replace, xsize←xnewPos-xstart, repSize])#
NIL
len ← len+x.oldPos-xstart; oldPos ← start+len;
repSize ← xsize+repSize; newPos ← start+repSize;
replace ← new; base ← x.base; LOOP
}
ELSE IF start+len = xstart AND -- replacing just before prior replacement
(new ← TryFlatConcat[replace, x.replace, repSize, xsize←xnewPos-xstart])#
NIL
THEN {
len ← len+x.oldPos-xstart; oldPos ← start+len;
 
repSize ← xsize+repSize; newPos ← start+repSize;
replace ← new; base ← x.base; LOOP
}
};
x: 
REF RunsBody.node.concat => {
xpos: INT ← x.pos;
IF start=xpos 
AND len=0 
THEN { 
-- insert between base&rest
first try concat of x.base&replacement
IF (new ← TryFlatConcat[x.base, replace, xpos, repSize])#
NIL
THEN 
RETURN [
NEW[Tconcat ←
[node[concat[size, new, x.rest, xpos+repSize]]]]];
 
 
otherwise try concat of replacement&x.rest
IF (new ← TryFlatConcat[replace, x.rest, repSize, x.size-xpos])#
NIL
THEN 
RETURN [
NEW[Tconcat ←
[node[concat[size, x.base, new, x.pos]]]]]
 
 
}
 
};
ENDCASE => NULL;
EXIT;
ENDLOOP;
 
IF base=NIL THEN base ← MakeRun[baseSize];
IF replace=NIL THEN replace ← MakeRun[repSize];
RETURN [NEW[Treplace ← [node[replace[size, base, replace, start, oldPos, newPos]]]]];
};
 
Copy: 
PUBLIC 
PROC [
dest: Runs, destLoc: INT, source: Runs, 
start, len, destSize: INT]
 
RETURNS [Runs] = {
merge: BOOL ← FALSE;
flat: BaseRuns;
looks: Looks;
c, numRuns: INT;
split, numruns: NAT;
Count: 
PROC [base: Runs, start, len: 
INT] 
RETURNS [
INT] = {
IF len=0 THEN RETURN[numRuns];
[c, merge, looks] ← CountRuns[base, start, len, FlatMax-numRuns, merge, looks];
RETURN [numRuns←numRuns+c]
};
 
Extract: 
PROC [base: Runs, start, len: 
INT] = {
IF len > 0 THEN split ← ExtractRuns[flat, base, start, len, split] 
};
 
IF dest=NIL AND source=NIL THEN RETURN[NIL];
numRuns ← 0;
IF Count[dest, 0, destLoc] > FlatMax OR
Count[source, start, len] > FlatMax OR
Count[dest, destLoc, destSize-destLoc] > FlatMax 
THEN 
RETURN [
Replace[dest, destLoc, 0, Substr[source, start, len], destSize, len, FALSE]];
IF numRuns=0 THEN RETURN [NIL];
flat ← NewBase[numruns←numRuns]; split ← 0;
Extract[dest, 0, destLoc]; Extract[source, start, len];
Extract[dest, destLoc, destSize-destLoc];
IF split # numruns THEN ERROR;
IF flat[numruns-1].after # destSize+len THEN ERROR;
RETURN [flat];
};